SlideShare a Scribd company logo
Biologically Inspired Approaches For Locomotion
Anomaly Detection And Reconfiguration For
Walking Robots 1st Edition Bojan Jakimovski Auth
download
https://ebookbell.com/product/biologically-inspired-approaches-
for-locomotion-anomaly-detection-and-reconfiguration-for-walking-
robots-1st-edition-bojan-jakimovski-auth-2451800
Explore and download more ebooks at ebookbell.com
Here are some recommended products that we believe you will be
interested in. You can click the link to download.
Biologically Inspired Approaches To Advanced Information Technology
First International Workshop Bioadit 2004 Lausanne Switzerland January
2930 2004 Revised Selected Papers 1st Edition Jacqueline Signorini
https://ebookbell.com/product/biologically-inspired-approaches-to-
advanced-information-technology-first-international-workshop-
bioadit-2004-lausanne-switzerland-january-2930-2004-revised-selected-
papers-1st-edition-jacqueline-signorini-2330358
Biologically Inspired Approaches To Advanced Information Technology
Second International Workshop Bioadit 2006 Osaka Japan January 2627
2006 1st Edition James C Liao Auth
https://ebookbell.com/product/biologically-inspired-approaches-to-
advanced-information-technology-second-international-workshop-
bioadit-2006-osaka-japan-january-2627-2006-1st-edition-james-c-liao-
auth-4140854
Computational Intelligence In Multifeature Visual Pattern Recognition
Hand Posture And Face Recognition Using Biologically Inspired
Approaches 1st Edition Pramod Kumar Pisharady
https://ebookbell.com/product/computational-intelligence-in-
multifeature-visual-pattern-recognition-hand-posture-and-face-
recognition-using-biologically-inspired-approaches-1st-edition-pramod-
kumar-pisharady-4698730
Biologically Inspired Control Of Humanoid Robot Arms Robust And
Adaptive Approaches 1st Edition Adam Spiers
https://ebookbell.com/product/biologically-inspired-control-of-
humanoid-robot-arms-robust-and-adaptive-approaches-1st-edition-adam-
spiers-5483834
Evolutionary Optimization Algorithms Biologicallyinspired And
Populationbased Approaches To Computer Intelligence Dan Simon
https://ebookbell.com/product/evolutionary-optimization-algorithms-
biologicallyinspired-and-populationbased-approaches-to-computer-
intelligence-dan-simon-36520926
Biologically Inspired Cognitive Architectures 2010 Proceedings Of The
First Annual Meeting Of The Bica Society 1st Edition A V Samsonovich K
R Jhannsdttir A Chella
https://ebookbell.com/product/biologically-inspired-cognitive-
architectures-2010-proceedings-of-the-first-annual-meeting-of-the-
bica-society-1st-edition-a-v-samsonovich-k-r-jhannsdttir-a-
chella-51692978
Biologically Inspired Cooperative Computing Ifip 19th World Computer
Congress Tc 10 1st Ifip International Conference On Biologically
Inspired Cooperative In Information And Communication Technology 1st
Edition Yi Pan
https://ebookbell.com/product/biologically-inspired-cooperative-
computing-ifip-19th-world-computer-congress-tc-10-1st-ifip-
international-conference-on-biologically-inspired-cooperative-in-
information-and-communication-technology-1st-edition-yi-pan-2141834
Biologically Inspired Networking And Sensing Algorithms And
Architectures Pietro Li
https://ebookbell.com/product/biologically-inspired-networking-and-
sensing-algorithms-and-architectures-pietro-li-2490906
Biologicallyinspired Optimisation Methods Parallel Algorithms Systems
And Applications 1st Edition Kalyanmoy Deb Auth
https://ebookbell.com/product/biologicallyinspired-optimisation-
methods-parallel-algorithms-systems-and-applications-1st-edition-
kalyanmoy-deb-auth-2543422
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
Cognitive Systems Monographs
Volume 14
Editors: Rüdiger Dillmann · Yoshihiko Nakamura · Stefan Schaal · David Vernon
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
Bojan Jakimovski
Biologically Inspired
Approaches for
Locomotion, Anomaly
Detection and
Reconfiguration for
Walking Robots
ABC
Rüdiger Dillmann, University of Karlsruhe, Faculty of Informatics, Institute of Anthropomatics,
Humanoids and Intelligence Systems Laboratories, Kaiserstr. 12, 76131 Karlsruhe, Germany
Yoshihiko Nakamura, Tokyo University Fac. Engineering, Dept. Mechano-Informatics, 7-3-1 Hongo,
Bukyo-ku Tokyo, 113-8656, Japan
Stefan Schaal, University of Southern California, Department Computer Science, Computational Learn-
ing & Motor Control Lab., Los Angeles, CA 90089-2905, USA
David Vernon, Khalifa University Department of Computer Engineering, PO Box 573, Sharjah, United
Arab Emirates
Author
Dr.-Ing. Bojan Jakimovski
Bionics4Robotics
Postfach 900609,
81506 München, Germany
E-mail: contact@bionics4robotics.com
ISBN 978-3-642-22504-8 e-ISBN 978-3-642-22505-5
DOI 10.1007/978-3-642-22505-5
Cognitive Systems Monographs ISSN 1867-4925
Library of Congress Control Number: 2011934501
c
 2011 Springer-Verlag Berlin Heidelberg
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9,
1965, in its current version, and permission for use must always be obtained from Springer. Violations
are liable for prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective
laws and regulations and therefore free for general use.
Typeset  Cover Design: Scientific Publishing Services Pvt. Ltd., Chennai, India.
Printed in acid-free paper
5 4 3 2 1 0
springer.com
Abstract
The increasing presence of mobile robots in our everyday lives introduces the re-
quirements for their intelligent and autonomous features. Therefore the next gen-
eration of mobile robots should be more self-capable, in respect to: increasing of
their functionality in unforeseen situations, decreasing of the human involvement
in their everyday operations and their maintenance; being robust; fault tolerant and
reliable in their operation.
Although mobile robotic systems have been a topic of research for decades and
aside the technology improvements nowadays, the subject on how to program
and making them more autonomous in their operations is still an open field for
research.
The classical formal methodologies which have been dominating the robotics
segment for long time start to prove that they are perhaps not adequate to cope
with the increasing complexity of the robotic systems. Many different research
directions have been considered on how to overcome these problems.
Applying bio-inspired, organic approaches in robotics domain is one of the
methodologies that are considered that would help on making the robots more
autonomous and self-capable, i.e. having properties such as: self-reconfiguration,
self-adaptation, self-optimization, etc.
In this book several novel biologically inspired approaches for walking robots
(multi-legged and humanoid) domain are introduced and elaborated.
They are related to self-organized and self-stabilized robot walking, anomaly
detection within robot systems using self-adaptation, and mitigating the faulty
robot conditions by self-reconfiguration of a multi-legged walking robot. The ap-
proaches presented have been practically evaluated in various test scenarios, the
results from the experiments are discussed in details and their practical usefulness
is validated.
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
Contents
1 Introduction ................................................................................................... 1
2 Biologically Inspired Computing and Self-x Properties............................. 5
2.1 Bionics..................................................................................................... 5
2.2 Organic Computing ................................................................................. 6
2.3 Autonomic Computing ............................................................................ 6
2.4 Self-x Properties ...................................................................................... 6
2.5 Emergence ............................................................................................... 7
3 Joint Leg Walking and Hybrid Robot Demonstrators............................... 9
3.1 Introduction ............................................................................................. 9
3.2 Hexapod Robots....................................................................................... 9
3.2.1 State of the Art – Hexapod Robots.............................................. 11
3.2.2 Hexapod Robot Demonstrator – OSCAR (Organic Self
Configuring and Adapting Robot) ............................................... 11
3.3 Humanoid Robots .................................................................................. 17
3.4 State of the Art Humanoid Robots......................................................... 17
3.5 Humanoid Robot Demonstrator - S2-HuRo
(Self Stabilizing Humanoid Robot) ....................................................... 19
4 Biologically Inspired Robot Control Architecture.................................... 23
4.1 Overview on “Standard” Types of Robot Control Architectures........... 24
4.1.1 Reactive and Subsumption and Behavior Based Control
Architecture ................................................................................ 24
4.1.2 Deliberative Control Architecture............................................... 25
4.1.3 Hybrid Control Architecture....................................................... 26
4.2 Overview on Autonomic Control Architecture...................................... 27
4.3 ORCA (Organic Robot Control Architecture)....................................... 29
4.4 Distributed ORCA Architecture for Hexapod Robot Control................ 30
4.5 Cell Differentiation as Biological Inspiration for Enhanced ORCA...... 31
4.5.1 Overview of a Biological Concept – Cell Differentiation........... 31
4.5.2 The Enhanced “Stem” Type ORCA Architecture....................... 32
5 Biologically Inspired Approaches for Locomotion of a Hexapod
Robot OSCAR.............................................................................................. 35
5.1 Characteristics of Locomotion Seen by Insects and
Animals - Applied to Robotics Domain................................................. 35
5.2 Central Pattern Generators (CPG) ......................................................... 37
VIII Contents
5.2.1 Common Observed Gaits by Insects........................................... 38
5.3 Experiments with Self-organizing Emergent Robot Walking
Gait with Distributed Pressure on Robot’s Feet..................................... 39
5.4 Firefly Inspired Synchronization of a Robot’s Walking Gait ................ 45
5.4.1 Firefly Coupled Oscillators Principle ......................................... 46
5.4.2 Concept for Robot Walking Gait Self-synchronization by
Using Firefly Synchronization.................................................... 47
5.5 Implementation of Firefly Inspired Self-synchronization into the
Robot Control Architecture ................................................................... 55
5.6 Experiments Done with Firefly Inspired Self-synchronization and
Results from Experiments...................................................................... 56
5.6.1 Experiment about Self-synchronization by Prolongation of the
Robot’s Swing and Stance Phases .............................................. 57
5.6.2 Experiment about Self-synchronization by Shortening of the
Robot’s Swing and Stance Phases .............................................. 60
5.6.3 Experiment about Self-synchronization by Combined
Prolongation and Shortening of the Robot’s Swing
and Stance Phases ....................................................................... 62
5.6.4 Discussion on Future Possible Improvements of Firefly
Inspired Self-synchronization Approach .................................... 66
5.6.5 Summary about the Firefly Inspired Self-synchronization
Approach..................................................................................... 66
6 Biologically Inspired Approach for Optimizing the Walking Gait of a
Humanoid Robot.......................................................................................... 67
6.1 Approaches for Walking Gait Generation by Humanoid Robots........... 67
6.2 Symbiosis as a Biologically Inspired Approach for
Self-stabilization of Humanoid Robot Walking Gait............................. 69
6.3 SelSta Approach in Detail...................................................................... 71
6.3.1 S2-HuRo Humanoid Robot Platform and Sensors Used............. 71
6.3.2 Control of the Robot S2-HuRo.................................................... 72
6.3.3 Main Parts of SelSta Approach – SymbScore Value and
Genetic Algorithm ....................................................................... 74
6.3.4 Fuzzy Logic Computation of SymbScore Value......................... 78
6.3.5 Genetic Algorithm Details for the SelSta Approach ................... 84
6.3.6 Preparation for Experiments........................................................ 85
6.4 Experiments Done with the SelSta Approach........................................ 88
6.4.1 Experiments on a Soft Green Carpet........................................... 89
6.4.2 Experiments on a Medium Soft Orange Carpet .......................... 97
6.4.3 Experiments on a Hard Green Carpet ....................................... 106
6.4.4 Experiments on a Hard Linoleum Surface................................ 115
6.5 Summary for Experiments Done with the SelSta Approach................ 124
7 Biologically Inspired Approaches for Anomaly Detection within a
Robotic System........................................................................................... 127
7.1 Overview on Approaches for Fault / Anomaly Detection by
Robotic Systems .................................................................................. 127
Contents IX
7.2 Overview of Artificial Immune System (AIS) Concept ...................... 128
7.3 Artificial Immune System Based - Robot Anomaly Detection
Engine (RADE) Approach................................................................... 131
7.3.1 Core Functionality of RADE Approach.................................... 134
7.4 Experiments Done with AIS Inspired RADE and Results from
Experiments......................................................................................... 135
7.4.1 Test-Bed Setup for RADE Approach........................................ 135
7.4.2 Self and Non-self Rule Sets by RADE ..................................... 136
7.4.3 Results from Experiments Done with the RADE Approach..... 138
7.4.4 3D representation of Run-Time Dynamics by RADE
Anomaly Detection Surface....................................................... 147
7.4.5 Summary about AIS Based Anomaly Detection
Approach - RADE .................................................................... 150
8 Approach for Robot Self-reconfiguration after Anomaly
Detection within a Walking Robot System Based on Biological
Inspiration - Swarm Intelligence............................................................. 151
8.1 Overview on Swarm Intelligence – Flocking Behavior and Boids...... 152
8.2 S.I.R.R. – Swarm Intelligence Based Approach for Robot
Reconfiguration ................................................................................... 155
8.2.1 Simulation of S.I.R.R Based Hexapod Robot
Reconfiguration ........................................................................ 158
8.3 Results from Robot Reconfiguration Experiments Done with
S.I.R.R. Approach on the Hexapod Robot OSCAR-2 ......................... 161
8.4 Results from Real Robot Reconfiguration Experiments Done with
S.I.R.R. Approach and Leg Amputations on the Robot OSCAR-X .... 162
8.4.1 Ground Contacts of Robot Legs for Normal Walking and for
Walking with Leg Amputations and Robot
Self-reconfiguration................................................................. 164
8.4.2 Tracking of the Robot’s Heading While the Robot Is
Performing Self-reconfiguration with Leg Amputations .......... 166
8.5 Summary for the S.I.R.R. - Biologically Inspired Robot
Reconfiguration Approach................................................................... 173
9 Conclusion and Outlook............................................................................ 175
10 References................................................................................................. 179
A Appendix.................................................................................................... 189
A.1 Test Bed for Tracking the Robot OSCAR-X during the
Experiments ....................................................................................... 189
List of figures .................................................................................................... 191
Keywords........................................................................................................... 199
Glossary............................................................................................................. 201
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 1–3.
springerlink.com © Springer-Verlag Berlin Heidelberg 2011
Chapter 1
Introduction
Robotic systems nowadays are getting increasingly complex in their design and
implementation. In order to fulfill the proposed requirements, systems often con-
sist of many software and hardware units, realizing various functionalities and
cooperating together. Declaring them as autonomous means that they should
have the ability to dynamically adjust and execute their tasks without human
intervention. Additionally, they should be reliable and also tolerant to various
system malfunctions.
Future robotic systems should be able to demonstrate self-x properties such as
self-organization, self-reconfiguration, self-healing and the like. Having these
kinds of properties, robotic systems would be able to demonstrate their autonomic
property, and this will also aid in shortening their development and maintenance
time.
However, the complexity of the classical approaches for robot system modeling
has introduced a need for developing and applying new concepts and methodolo-
gies towards creating self-capable, more robust, and dependable systems. In order
to achieve this, engineers have used different biologically and organically inspired
approaches. For example, some of the algorithms for motor control and walking
gait pattern generations for the domain of joint leg walking robots have been
inspired and developed on observations seen within animals and functioning of the
neural circuitry. In that context, research has been done and is presented in this
book on introducing novel biologically inspired approaches with their practical
applications for domain of self-organizing, self-optimizing, and self-reconfiguring
walking robots. Joint leg walking robots and simple walking robots use their legs
for their movement over a terrain. Depending on how many legs they have, they
can belong to one of many categories, from two legged humanoid robots up to
many legged robots having three, four, six, or eight legs, where each leg can be
built out of several joints/segments. Walking robots can have many joints and
therefore many degrees of freedom (DOF). However, having many degrees of
freedom introduces difficulties in developing appropriate methods for controlling
such complex walking robots, monitoring their health status, making them fault-
tolerant, etc.
The research presented in this book tries to give an answer on some of the
open questions for fault-tolerant robotics domain. This includes self-organized,
2 1 Introduction
self-stabilized robot walking and self-adapting methods for failure detection and
self-reconfiguration after a failure has been detected within the robotic system, so
the robotic system can still continue with its mission despite the faulty conditions
within itself.
Research presented in this book includes:
- biologically inspired robot control architecture;
- self-organizing hexapod robot walking;
- biologically inspired self-stabilizing humanoid robot walking;
- biologically inspired self-adapting approach for anomaly detection
by a robot;
- biologically inspired approach for self-reconfiguration by a hexapod
robot.
The structure of the thesis is organized as follows:
- In the 2nd
chapter, a general overview is given on biologically inspired com-
puting and self-x properties including short introduction to terms related to:
bionics, organic computing, autonomic computing, self-x properties, and
emergence;
- The 3rd
chapter is about joint leg walking and hybrid robots, with a small
review about state of the art of hexapod robots. In this chapter detailed de-
scriptions about the hexapod robot demonstrators OSCAR-2, OSCAR-3,
and OSCAR-X are given. State of the art humanoid robots are further in-
troduced and also a detailed description of the S2-HuRo humanoid robot
demonstrator is given.
- The 4th
chapter describes notions about a biologically inspired robot con-
trol architecture. An overview is given on commonly used robot control
architectures such as: reactive, subsumption, deliberative, and hybrid con-
trol. After the introduction on commonly used robot control architectures,
an introduction is given on self-organizing robot architectures such as:
autonomic control architecture, Organic Robot Control Architecture
(ORCA), and their characteristics. The distributed ORCA which is related
to the research experiments is explained in detail and new ideas are also
given about an enhanced “stem” based ORCA architecture.
- The 5th
chapter first gives an introduction on locomotion seen by insects
and animals and Central Pattern Generator (CPG) for walking pattern gen-
eration. Then a concept for self-organizing emergent robot walking gait
with distributed pressure on robot’s feet is explained. Further explained in
this chapter is the firefly-inspired synchronization of a robot’s walking
gait. Included are the experiments done on prolongation and shortening of
the robot’s swing and stance phases using firefly-inspired synchronization.
- In the 6th
chapter a biologically inspired approach for optimizing the walk-
ing gait of a humanoid robot is explained. Symbiosis as a biologically
inspired approach for self-stabilization of humanoid robot walking gait is
elaborated and details about the SelSta approach and its main parts are
Introduction 3
discussed. The chapter concludes with the results of experiments done on
the SelSta approach and the usefulness of the SelSta approach.
- The 7th
chapter is about biologically inspired approaches for failure detec-
tion within a robotic system. First it gives an overview on the approaches
for robot fault / anomaly detection, followed by an introduction on Artificial
Immune Systems. Then it introduces the Artificial Immune System (AIS)
based Robot Anomaly Detection Engine (RADE) approach. After RADE is
explained in more detail, results from the experiments done with the AIS-
inspired RADE approach are presented and discussed.
- The 8th
chapter introduces an approach for robot self-reconfiguration of a
hexapod robot system based on an biological inspiration - swarm intelli-
gence. An overview of swarm intelligence, flocking behavior, and boids is
given first. Then S.I.R.R., a Swarm intelligence based approach for robot
reconfiguration, is introduced and explained in detail. After the introduc-
tion of the S.I.R.R. approach, results from simulation of S.I.R.R.-based
hexapod robot reconfiguration are presented. The results from real robot
reconfiguration experiments done with the S.I.R.R. approach on the hexa-
pod robot OSCAR-2 are then presented for validation of the simulation
experiment’s results. Chapter 8 ends with the results from real robot recon-
figuration experiments done with the S.I.R.R. approach, presentation of leg
amputations on robot OSCAR-X, and an explanation of practical useful-
ness of S.I.R.R.
- The 9th
chapter gives a conclusion on the research presented in this book
and the importance of the biologically-inspired approaches introduced for
the walking and general robotics domain.
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 5–7.
springerlink.com © Springer-Verlag Berlin Heidelberg 2011
Chapter 2
Biologically Inspired Computing and Self-x
Properties
2.1 Bionics
Bionics is the application of biological methods and systems found in nature to the
study and design of engineering systems and modern technology [Bio10].
The term Bionics (from biology and electronics) is sometimes interchangeably
used for a Biomimetics and Biomimicry (from bios = life, and mimesis = to imi-
tate). Bionics is related to applying ideas seen in nature for solving scientific,
technical, or engineering problems. Biomimetics is therefore an interdisciplinary
field where scientists from different scientific fields like Biology, Physics, Chem-
istry, and Engineering work together towards developing various solutions based
on observations of processes seen in nature. All these techniques are based on
solving new problems from solutions to previous problems found in nature.
For example, such biologically inspired concepts can be related to various
organizational building principles seen within bacteria, flora, fauna, etc.
There are many innovations and products developed that can be mentioned as
examples for practical usefulness of biologically inspired concepts: self-
assembling glass inspired by sea sponges; bacterial control inspired by red algae;
solar cells inspired by leaves; friction-free fans inspired by nautilus; building ma-
terial from CO2 inspired by mollusks; self-cleaning surfaces inspired by lotus
plant [Nat10], etc.
Besides the natural assembling techniques which can be useful for building
novel materials with new and perhaps superior properties than that of the current
existing materials, the social and organizational principles in nature like feeding,
mating, foraging, and swarming are also found to be useful for engineering
domain.
Bionics is more related to implementing the approach found in nature as an idea
instead of imitating the biologically structure behind it, which is more closely
representing the Biomimetics.
Nowadays there is an increasing trend of applying and adopting biologically in-
spired approaches for the general domain of computer science as well as for the
domain of robotics.
6 2 Biologically Inspired Computing and Self-x Properties
More and more of the approaches implemented in the field of computer engi-
neering are related to artificial intelligence and optimization, such as artificial neu-
ral networks, genetic algorithms, and ant optimization algorithms.
2.2 Organic Computing
Organic computing initiative [Mül04] [Sch05], is related to development of tech-
nical systems that act autonomously and dynamically adapt to the environment.
Those systems must exhibit lifelike properties and function in independent way.
That is why they are called “organic” or “organic computing systems”. At the
same time they should be also robust, safe, and trustworthy.
Therefore, organic computing is a type of biologically inspired computing that
attempts to develop approaches for technical systems exhibiting self-x properties
such as self-organization, self-configuration, self-optimization, self-healing, and
self-explaining.
The research presented in this book is directly associated with Organic Com-
puting and developing self-x biologically inspired approaches that would enable
the robotic systems to act in more independent and autonomous ways.
2.3 Autonomic Computing
The Autonomic computing initiative was proposed in the IBM [IBM01] manifesto
and states the need for development of autonomic IT systems that would over-
come the ever growing complexity of current IT systems. The main requirement
for such autonomic systems would be that they are self-manageable and also
capable of providing reliable services and minimizing the human administrator
intervention and thus minimizing the probability of human errors.
Autonomic systems are therefore systems that can manage themselves without
human intervention. They must be capable of incessant autonomous work given
only high-level objectives from the administrators [KeC03]. In order to achieve
the autonomic system's property, such systems must have the self-x properties
self-reconfiguration, self-organization, and self-healing.
2.4 Self-x Properties
Self-x properties are closely related to biological processes found in nature,
namely the processes in nature that can be often seen as self-organizing, self-
optimizing, and self-healing processes.
Such self-x properties have been proven useful when translated to scientific
domains: techniques for new materials development, engineering, new approaches
for the IT industry, etc.
Approaches mentioned in this book are mainly related to the development of
various algorithms for joint leg walking robots domain that enable the robots to
2.5 Emergence 7
exhibit the so-called self-x properties for various circumstances such as robot self-
reconfiguration, robot walking gait self-optimization, and robot self-healing.
General definitions for the terms Self-Configuration, Self-Optimization, Self-
Healing, and Self-Protection can be found by Autonomic Computing [KeC03] and
also can be used interchangeably for other technical domains, as well as for the
robotics domain.
Self-Configuration: Automated configuration of components and systems follows
high-level policies. The rest of the system adjusts automatically and seamlessly.
Self-Optimization: Components and systems continually seek opportunities to
improve their own performance and efficiency.
Self-Healing: The system automatically detects, diagnoses, and repairs localized
software and hardware problems.
Self-Protection: The system uses early warning to anticipate and prevent system-
wide failures.
2.5 Emergence
Emergence is one of the phenomena often observed in nature. Emergence can be
defined as “the arising of novel and coherent structures, patterns, and properties
during the process of self-organization in complex systems [Gol10]. The emer-
gence can sometimes be summarized as: Whole is more the than sum of its parts.
This means that in systems exhibiting emergence, the behavior or the property of
the whole system cannot be deducted from the properties of individual compo-
nents composing that system. Such a definition is closer to the view of “strong
emergence”. On the other side, “weak emergence” is related to the emergence that
is traceable, i.e. the emergent property can be reduced to the property of individual
components.
For emergence is often said to be a “bottom-up” process.
There are many examples of emergent processes that can be seen in nature or in
biological systems. For example: the sand dunes, water waves, swarming schools
of fish, flocking of birds, slime molds, ant colonies’ self-sustainability, etc.
In complex systems where safety is not a critical issue (since the completely
safe system's behavior cannot be guaranteed), emergence is sometimes used to
lower the effort of developing the needed system's functionality.
Emergence is also popular for domain of robotic systems. Examples include:
emergence of gait patterns for robot walking [AWY99], emergence of communica-
tion within multi robot systems [Lip07], emergent behaviors of autonomous robots
[AnD90], etc.
The practical usefulness of the emergence concepts is explained in further
chapters, applied to the research on the hexapod robot OSCAR, more precisely for
its walking gait generation.
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 9–21.
springerlink.com © Springer-Verlag Berlin Heidelberg 2011
Chapter 3
Joint Leg Walking and Hybrid Robot
Demonstrators
3.1 Introduction
Design of two and multi-legged robots like four legged, six legged, and eight
legged robots shows a practical usefulness of Bionics for the domain of robotics.
Depending on the number of legs, such robotic platforms can be inspired from
body constitution, walking mechanics, and behavior of humans, animals (four
legged), insects (six legged), or spiders (eight legged). There is also another kind
of robotic designs depicted as hybrid robots, which have a mixture of concepts
seen in nature and artificial designs. These robots can be often found having
tracks, other special leg designs, wheeled-legged robot designs and the like.
In this chapter several joint leg walking and hybrid robots are discussed. The
“joint leg” term is related to robots that have legs built out of servos, representing
their joints. The term “hybrid” found by hybrid robots is related to robots that
have both legs built out of servo joints and wheels attached to their legs.
Robots demonstrators presented in this chapter have been used as demonstra-
tors for biologically inspired approaches and algorithms researched and elaborated
in this book. These robots can be categorized into three types: humanoid robots,
hexapod-robots, and hybrid wheeled-legged robots.
3.2 Hexapod Robots
Hexapod robots belong to the group of joint leg walking robots having six legs
where the legs are consisting of multiple servo joints. The legs of the robot are
usually symmetrically distributed in two different groups spatially located on the
two opposite sides of the robot's body. The design of hexapod robots is often in-
spired by locomotion systems seen in insects like cockroaches, stick insects, and
the like.
In comparison with the four legged walking robots or quadruped robots,
hexapod robots have intrinsically more redundancy due to the higher number of
the legs and thus can be theoretically more flexible over uneven terrain. Hexapod
10 3 Joint Leg Walking and Hybrid Robot Demonstrators
robots differ from robots that have “native” spider-like biomimetic design having
eight legs distributed on the two sides of the robot’s body. Although the eight
legged robots may have higher degree of redundancy and perhaps provide better
agility for the robot over rugged terrain, they also need more energy for their func-
tioning, which in turn affects the size and mobility of the robot.
(a) (b)
(c) (d)
(e)
Fig. 3.1 Hexapod robots: (a) “iSprawl”; (b) “RHex”; (c) “DLR Crawler”; (d) “RiSE”; (e)
“AMOS-WD06”.
3.2 Hexapod Robots 11
3.2.1 State of the Art – Hexapod Robots
There have been many types of hexapod robots that have been used for demon-
stration purposes in the research on biologically inspired locomotion. Some of the
current (Feb, 2010) state of the art hexapod robots include: “iSprawl” [KCC06],
“RHex” [AMK01], “DLR Crawler” [GWH09], “RiSE” robot [SGF06], “AMOS-
WD06” [STW10] (Figure 3.1). However, the mentioned hexapod robots all differ
in the technology that they use for their locomotion. For example, their leg design
differs from one robot to the other and the moving concept of the joints by the legs
also differs.
Movement of the legs for the “iSprawl” robot is periodic, generated by push-
pull actions using flexible cables and servo motors [KCC06]. For the “RHex”
robot, the movement of the legs is related only to the rotary motion of the legs
[AMK01]. The “DLR Crawler” design was based on the “DLR-HAND II”
[BFH03], therefore the joint based fingers of the “DLR-HAND II” are adapted to
serve as legs for the “DLR Crawler” [GWH09]. For the “RiSE” robot, the legs are
moved by two electrical actuators per leg, with biologically inspired adhesive
structures located on the feet, which enable the “RiSE” robot to climb on vertical
wall surfaces, trees, etc [SCM06]. The “AMOS-WD06” robot implements legs
consisting of three servos each, resulting in eighteen servos for locomotion of the
hexapod robot [STW10].
By the presented state-of-the-art robots there are improvements that can be
seen in comparison with some older hexapod robot designs, however less has been
done on introducing fault tolerant mechanisms within the robots itself, which
will give the robots the robustness to be functional also in situations when they
experience some malfunctions within their components, by means of reconfigur-
ing the body and leg postures.
And this is the key difference in comparison to the robot demonstrators OSCAR,
in particular OSCAR-X, which were developed at Institute für Technische Infor-
matik, University Lübeck and described in the following sub-chapter.
3.2.2 Hexapod Robot Demonstrator – OSCAR (Organic Self
Configuring and Adapting Robot)
OSCAR (Organic Self Configuring and Adapting Robot) is a six legged walking
robot, used as a demonstrator for testing some of the newly-developed biologically
inspired approaches and algorithms presented in this book.
OSCAR represents the series of built hexapod robot demonstrators used in the
interdisciplinary research, where the legs are distributed spatially in a circle on the
robot's body. Most of the robots in OSCAR series have integrated an on-board
embedded system, sensors (ultrasonic, infrared, acceleration, and inclination sen-
sors), and actuators (analog and digital servos). Such a typical spatial distribution
of the robot’s legs in a circle is represented in (Figure 3.2).
12 3 Joint Leg Walking and Hybrid Robot Demonstrators
Fig. 3.2 Spatial distribution in circle of legs by OSCAR series of robots.
The OSCAR series consists of the following robots: OSCAR-1, OSCAR-2,
OSCAR-3, and OSCAR-X, which are described below. The first two robots,
OSCAR-1 and OSCAR-2, were mostly based on the Lynxmotion® robot kit -
“AH3-R (18 Servo Walker)” with 18 degrees of freedom (DOF), Hitec HS-645
Servos and aluminum based leg design [Lyn06].
3.2.2.1 Hexapod Robot Demonstrator – OSCAR - 1
Hexapod robot OSCAR-1, built in year 2006 is the first in the series of OSCAR
robots. Its hardware is based on the “AH3-R (18 Servo Walker)” robot kit with six
legs distributed spatially in a circle and additional on-board electronics such as
“JControl” (a Java based embedded system for robot control), servo controller SD-
21, Hitec analog servos HS-645, binary contact sensors on the robot's feet, and
NiMH batteries (Figure 3.3). Each leg by the robot is made up of three servos.
There are also integrated ultrasonic sensors on three of the robot’s legs.
Fig. 3.3 Hexapod robot OSCAR-1
3.2 Hexapod Robots 13
3.2.2.2 Hexapod Robot Demonstrator – OSCAR - 2
OSCAR-2 (Figure 3.4) is the second in the series of OSCAR robots, similarly
built as OSCAR-1.
The OSCAR-2 in comparison to the OSCAR-1, has the following modifications:
- pressure sensors (Figure 3.5);
- 18 modified HiTec HS-645 servos (Figure 3.6);
The modified servos provide feedback for the level of servo current, so the torque
can be monitored while the robot is walking. The modification is clearly visible due
to the number of wires that come out from the servos, namely that the wires are
directly connected to the potentiometer output reading pins inside of the servos.
(a) (b)
Fig. 3.4 Hexapod robot OSCAR-2. (a) Experimental robot OSCAR-2 setup - from above;
(b) Robot OSCAR-2 in movement.
Fig. 3.5 Pressure sensors type FSR-400. The most right one in the figure is used by
OSCAR 2.
Another difference to OSCAR-1 is that OSCAR-2 has pressure sensors on its
feet (Figure 3.5) instead of the binary contact sensors, so a variable pressure on the
robot’s feet can be sensed.
Furthermore, OSCAR 2 has an accelerometer sensor used to sense the accelera-
tion and inclination of the robot.
14 3 Joint Leg Walking and Hybrid Robot Demonstrators
By experiments with OSCAR-2, National Instruments hardware [Nat06] and
software was used for acquisition and pre-processing of the signals (currents from
servos, feet pressure, inclination values), their graphical representation, and data
logging.
Fig. 3.6 Modified HiTec HS-645 servo with wires for current and position feedback
Other important characteristic for robot OSCAR-2 is that in order to simulate a
faulty situation of the robot’s legs by anomaly detection experiments, there have
been modifications introduced to some of the robot’s legs for those particular ex-
periments. Such experimental leg modification is shown in (Figure 3.7).
Fig. 3.7 Modification by leg of robot OSCAR-2, in order to allow simulated leg failure
The presented modification allows the robot’s leg to intentionally malfunction
(the inserted pins drop off) after some time of robot walking.
3.2.2.3 Hexapod Robot Demonstrator – OSCAR - 3
OSCAR-3 is similar to OSCAR-1 and OSCAR-2 in its principal construction. The
difference to OSCAR-1 and OSCAR-2 is that the 18 modified HiTec HS-645 ser-
vos have additional internal electronic printed board circuits that provide servo
current feedback through an I2C bus back to the computing unit. Robot OSCAR-3
3.2 Hexapod Robots 15
doesn’t have a microcontroller onboard, but instead it is connected to a PC via a
USB cable. It uses the “Generic robot architecture” [Gen08] concept in order to
provide a better software driver access to the robot’s sensors and actuators, and
therefore easier control and actuation of the robot.
Fig. 3.8 Hexapod robot OSCAR-3.
3.2.2.4 Hexapod Robot Demonstrator – OSCAR - X
The new prototype of the OSCAR robot generation, called OSCAR-X (Figure 3.9),
is built to provide a better robot research test-bed for testing the biologically in-
spired algorithms. In comparison to its predecessors, the OSCAR-X features a
completely new design and was rebuilt from scratch.
New features of the robot include:
- Robot leg amputation mechanism: R-LEGAM [Jak09];
- Light weight glass-fiber body;
- Robot legs spatially distributed in a circle with 60 degrees between
each two neighboring legs;
- Greater payload capabilities (sensors, batteries, camera, etc.) for the
scientific measurements and experiments;
- Stronger digital RX-64 servos with digital feedback for their real time posi-
tions, torque levels, current levels, temperatures, etc.
- Powerful Lithium-polymer batteries for the servos and electronics;
- Weight of the body including the batteries is 7,5 kg;
- Improved foot design for better detection of the ground, complete with binary
contact sensors;
- Powerful embedded system - Gumstix® “Verdex board” [GUM09] running
embedded Linux;
- Usage of the “Generic robotic architecture” [Gen08] concept, to provide bet-
ter software driver access to the robot’s sensors and actuators and therefore
easier control and actuation of the robot;
- Orientation sensor;
- Wireless camera and an additional camera servo.
16 3 Joint Leg Walking and Hybrid Robot Demonstrators
(a) (b)
(c) (d)
Fig. 3.9 (a) Hexapod robot OSCAR-X in development stage; (b) OSCAR-X in nature; (c)
Front view of robot OSCAR-X with onboard camera and additional ultrasonic sensors; (d)
Top view of robot OSCAR-X.
3.2.2.4.1 Robot Leg Amputation Mechanism – R-LEGAM
The main feature of the OSCAR-X is the improved design of robot’s legs, which
aim for performing on-demand robot reconfiguration. Namely, the patented
mechanism for robot leg amputation, R-LEGAM (DPMA-Az: 10 2009 006 934)
[Jak09], is integrated for each of the OSCAR-X’s legs (Figure 3.9 and Figure 3.10).
The robot’s leg can be detached from the robot’s body by software command.
This is especially helpful when some of the legs malfunction. So instead of carry-
ing the malfunctioned legs during the rest of the mission, the legs can be ampu-
tated to prevent any other future negative influence on the rest of the functional
robotic system.
The in-situ reconfiguration of the hexapod robot OSCAR-X using biologically
inspired approaches will be discussed in chapter 10.
3.3 Humanoid Robots 17
Fig. 3.10 (a) CAD design of Robot leg amputation mechanism: R-LEGAM; (b) R-LEGAM
integrated on the robot’s body; (c) Robot’s leg detached from the robot’s body using the R-
LEGAM mechanism.
3.3 Humanoid Robots
Humanoid robots are robot demonstrators that have a human like appearance and
are often used in robotics research, or as entertainment and service robots.
Humanoid robots are therefore used to study and research the complexity of
human walking and dynamic balancing, but also used for research in prosthesis
development, human cognition, and human sensory information processing and
perception.
They have two legs, usually two arms and a head, and are equipped with lots of
actuators and sensors including accelerometers, tilt sensors, cameras, pressure
sensors on their feet, ultrasonic and infra-red sensors, etc.
There are humanoid robot soccer matches organized by the RoboCup federa-
tion [Rob10], where humanoid robots autonomously play soccer. Such competi-
tions are important for the overall research in humanoid robots and emphasize the
work on developing new algorithms for humanoid robot dynamic walking and
stabilization, cooperation, localization on the field, etc.
In chapter 6, research work is presented on self-stabilizing humanoid robot
walking using biologically inspired algorithms.
3.4 State of the Art Humanoid Robots
There are many humanoid robots known nowadays (March 2010) that are con-
sidered state-of-the-art due to the number of features they exhibit. Here are some
of the famous humanoid robots: Nasa’s “Robonaut 2” (R2) [NAS10], “TOPIO
3.0” [TOS09], “ASIMO” developed by HONDA [HON07], “Albert-Hubo”
[HAN05] by Hanson Robotics, and “NAO” by Aldebaran Robotics [ALD10].
(Figure 3.11)
18 3 Joint Leg Walking and Hybrid Robot Demonstrators
(a) (b)
(c) (d) (e)
Fig. 3.11 State of the art humanoid robots: (a) “Robonaut 2”; (b) “TOPIO 3.0”; (c)
“ASIMO”; (d) “Albert-Hubo”; (e) “NAO”.
The robot “Robonaut2,” nicknamed as “R2,” is a dexterous and technologically
advanced humanoid robot developed by NASA and General Motors. The goal is
to have this robot accompany the space missions and work side-by-side with
humans. The “R2” has a torso equipped with a head and two arms but is without
legs.
“TOPIO 3.0” is the table tennis playing robot, designed and constantly im-
proved by the company Tosio. The robot is said to use artificial intelligence algo-
rithms to continuously improve its playing skill level.
Robot “ASIMO,” developed by HONDA, is one of the most famous humanoid
robots and is mostly used for entertainment purposes. This robot can detect faces,
shake hands with humans, walk up and down the stairs, run, and even perform
small jumps while running.
Robot “Albert-Hubo” is built on “Hubo 2,” a “KHR-4” [HUB03] robot model,
and is the next generation of the “KHR-3” humanoid robot with Albert Einstein’s
head mounted on its body. The robot is used for artificial muscle actuator re-
search, autism therapy, cognitive science research, etc.
3.5 Humanoid Robot Demonstrator - S2-HuRo (Self Stabilizing Humanoid Robot) 19
“NAO” is commercially available research humanoid robot platform equipped
with a variety of actuators and sensors like: cameras, gyrometers, accelerometers,
IR, and sonar sensors. It has a visual programming interface with which robotic
movements can easily be developed. “NAO” robots are also in the RoboCup hu-
manoid soccer games in the NAO - RoboCup standard league.
Humanoid robots come in various sizes, ranging from small robots like NAO
robot - 58 cm up to full size robots like TOPIO 3.0 - 188cm size (Figure 3.11).
3.5 Humanoid Robot Demonstrator - S2-HuRo (Self Stabilizing
Humanoid Robot)
The S2-HuRo robot was developed for the research on biologically inspired tech-
niques for humanoid robot stabilized walking, presented in chapter 6. S2-HuRo is
based on the “ROBONOVA-1” (Figure 3.12) [HIT06] robot kit with “HSR-
8498HB” digital servos, “MRC-3024” servo control, and integrated I/O board.
Fig. 3.12 Humanoid robot “ROBONOVA-1”
This robot has been modified extensively and specially tuned, taking the final
form and construction as shown in Figure 3.12.
The modified robot is called the S2-HuRo (Self Stabilizing Humanoid Robot)
presented in Figure 3.13 (a)-(d). The robot is built of the following hardware ele-
ments: embedded system - Gumstix® Verdex [GUM09] represented in Figure
3.14 with wireless module and antenna, 2-axis accelerometer and gyroscope, Lith-
ium-polymer battery pack, voltage converters for 3, 5, and 6 volts for the electron-
ics, and binary sensor contacts (Figure 3.15). In order to decrease the weight of
the robot, some servos from the “Robonova 1” arms have been removed during
the robot hardware tuning, and the connection between the segments at those
points is made with screws instead. The final S2-HuRo robot appears as seen in
Figure 3.13.
20 3 Joint Leg Walking and Hybrid Robot Demonstrators
(a) (b) (c)
(d) (e)
Fig. 3.13 (a) – (d) S2-HuRo (Self Stabilizing Humanoid Robot).
Fig. 3.14 Gumstix® Verdex embedded system with wireless LAN module, antenna, MMC
card, and additional serial connector cable. View with three sensors per foot. “L” and “R”
indicate the left and the right robot’s legs.
3.5 Humanoid Robot Demonstrator - S2-HuRo (Self Stabilizing Humanoid Robot) 21
When comparing the robots shown in Figure 3.12 and Figure 3.13, differences
in robot modifications can easily be spotted. By looking at the top view of the ro-
bots, the robot (Figure 3.13) (c) the two axis accelerometer sensors can be spotted
under the robot’s head. In the backside figures (Figure 3.13) (d) and (e) of the S2-
HuRo, the case of the embedded Gumstix® Verdex system can be seen along with
the wireless antenna next to the case.
On the robot’s feet Lithium-polymer batteries can be spotted – two batteries per
foot. The relatively light weight 5Wh LiION rechargeable batteries used by
S2-HuRo for servo and electronics power supply are the same batteries used by
E-pucks robots [EPU09]. The voltage from the batteries is further down-regulated
by integrated voltage convertors to be compatible with the voltage operating range
of the servos, Microcontroller AtMega “MRC-3024” board, and the additional
electronics. The batteries are located on the robot’s feet so the center of gravity of
the robot is lowered and this increases the dynamic stability of the robot.
Fig. 3.15 Schematic view of binary contact sensors by the S2-HuRo feet – bottom.
Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth
B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 23–34.
springerlink.com © Springer-Verlag Berlin Heidelberg 2011
Chapter 4
Biologically Inspired Robot Control
Architecture
Robot control architectures are related to sensing, monitoring, and acting actions
of the robots. They are an important part of each robot control and the coordina-
tion of their behaviors. There are different kinds of robotic architectures imple-
mented by different kinds of robots.
Most of the control robot architectures are divided into following groups:
- Reactive and Subsumption / Behavior-based control architectures (schema
based);
- Deliberative control architectures (hierarchical) or Sense-plan-act architec-
tures;
- Hybrid control architectures.
These “common” types of robotic architectures are explained in the subsection
that follows.
On the other hand, the research of biologically inspired robot control architec-
tures aims at developing robot control architectures that would have life-like
properties and are able to self-organize their constitutive components instead of
predefining them manually. They must be able to self-optimize for their best
performance and be capable of detecting the newly attached components such as
sensors and actuators. They must also be able to autonomously re-configure them-
selves and continue with execution of their mission tasks.
Although these are all nice features to have for robot control architecture, de-
veloping such “life-like” robot control architecture is rather complicated. The
tricky design of such architecture involves re-thinking aspects such as:
- The types of elements that would constitute such control architecture;
- The property of the architecture must be generic and independent of the
configuration and attached hardware and electronics;
- How can an on-demand reconfiguration be achieved in such autonomic ar-
chitecture so that the new components can be easily added to the running
system?
- Where to place the “main control” unit or if such a unit should exist at all
in such module-based architecture.
24 4 Biologically Inspired Robot Control Architecture
Questions also arise on how the decision for a particular detailed design of such
architecture might influence the proper working of the other constitutive elements
of the architecture. Properties such as scalability and emergence also have to be
addressed.
4.1 Overview on “Standard” Types of Robot Control
Architectures
In this section several common types of robot control architectures are introduced:
Reactive control architecture; Subsumption control architecture; and Delibera-
tive/reactive control architecture. Then an overview on autonomic types of
architectures is given, before an explanation of the ORCA (Organic Robot Control
Architecture). ORCA is on the other side directly related to the research presented
in this book as well as to several ideas that are presented about improvement of
robot control architecture using biologically inspired concepts.
4.1.1 Reactive and Subsumption and Behavior Based Control
Architecture
By reactive control architectures (schema based), the control is stimulus-response
based, where behaviors are represented by direct sensor to actuator predefined
reaction mappings. (Figure 4.1) Although the speed of response by the reactive
control architecture is rather high, which might be suitable for some real world
scenarios where the reaction time might be very important, reactive architectures
are perhaps not suitable for tasks where predictive planned outcomes should be
generated.
Fig. 4.1 Principle of reactive control architecture.
An alternative approach to the reactive system architecture is subsumption
architecture (behavior based) introduced by Brooks in 1986 [BRO86]. This ap-
proach is based on priority behaviors organized into layers, where higher priority
behaviors subsume lower priority behaviors. In subsumption architecture, the low-
er layer behaviors (reflexes) can inhibit higher layer behaviors. In this bottom-up
fashion, the reflexes are bottom layers that can be expressed in the fastest way.
The upper layers are related to higher robot control work in inhibited fashion, de-
pending on their priority and currently executed lower actions. It is also important
to mention that there is no higher level supervision in this architecture. The
subsumption architecture can be useful when overall robot behavior should be
4.1 Overview on “Standard” Types of Robot Control Architectures 25
dynamic, reactive, emergent. Difficulties in this architecture are related to deter-
mining the priorities of the layers constituting the architecture. An example struc-
ture of subsumption architecture is presented in Figure 4.2.
Fig. 4.2 An example structure of subsumption architecture.
4.1.2 Deliberative Control Architecture
Deliberative control architectures are based on the Sense-Plan-Act principle. For
their optimal functioning, deliberative hierarchical control architectures usually
need full knowledge about the environment.
By these types of control architectures, the robot first senses the environment,
then plans potential solutions and considers the results when choosing appropriate
actions. It is assumed that the world model is provided. The robot then executes
the actions through actuators. The structure of such architecture is represented in
Figure 4.3.
Fig. 4.3 Sense-Plan-Act model of deliberative control architecture.
26 4 Biologically Inspired Robot Control Architecture
The advantages of using the deliberative architecture is that in such goal ori-
ented control architecture the goal of a given task can be achieved in a planned
way. However, the difficulties or drawbacks of using such architecture are related
to the re-planning phases which introduce slow response to some actions. There-
fore the architecture is perhaps not suitable for tasks where fast response is
needed. An additional drawback is that in case the environment changes, there
also need to be changes in the control architecture, so its reaction can be compati-
ble with the changed model of the environment.
4.1.3 Hybrid Control Architecture
One way of mitigating the limitations and drawbacks seen by the reactive and the
deliberative control architectures is to combine both of the architectures into a
hybrid control architecture.
A first proposal for usage of such hybrid architecture was made by Arkin
[Ark87]. Since then, different kinds of hybrid robot architectures have been pro-
posed [Con92] [LHG06]. In general, the hybrid architecture uses higher level
planning in order to guide the lower level of reactive components. It is often
depicted as a three layer architecture, where the top layer is the deliberative
layer, operating under a slower sampling rate than the bottom layer, which is the
reactive layer with a fast reaction time. The middle layer might have different
interpretations and implementations for different projects, for example, aggrega-
tion of information coming from the lower layer. This control architecture can be
represented as in Figure 4.4.
Deliberative layer
Intermediate layer
Reactive layer
Fig. 4.4 Model of a hybrid control architecture.
The benefit of using such architecture is that it is still a goal oriented architec-
ture where planning for the next actions occur by a deliberative layer, and at the
same time “lower level” actions can be executed by the reactive layer. Therefore it
can be more or less assured that the proper planned actions will be executed
4.2 Overview on Autonomic Control Architecture 27
and also that the robot will interact better and react faster to changes in the
environment.
4.2 Overview on Autonomic Control Architecture
Previously introduced robot control architectures (reactive, deliberative, and
hybrid architectures) have the modules, behaviors, and tasks mostly planned,
modeled, and defined in advance by a human operator. Autonomic control archi-
tectures use the idea to build architectures that will easily cope with the high
complexity of the technical systems, and that dynamically adapt with respect to
available resources and user needs [NaB07]. Responses taken automatically by a
system without real-time human intervention are called autonomic responses
[SHR06] [LVO01].
Given only high-level commands, the autonomic systems should be able to
manage themselves [KeC03] in a self-governing manner. The idea of autonomic
computing was first introduced by IBM in their Manifesto for Autonomic Com-
puting [IBM01]. They proposed several features that autonomic systems should
exhibit such as: self-configuration, self-healing, self-optimization, and self-
protection, all of which were inspired by the human body’s autonomic nervous
system. These terms were explained previously in Chapter 2.4.
Autonomic systems consist of autonomic elements – which can build relation-
ships with other autonomic elements and manage and influence or change their
behavior in order to comply with the higher level policies defined by human op-
erators. Such an autonomic element is represented in Figure 4.5.
As seen from the figure, each autonomic element has an autonomic manager
and one or more managed elements. An Autonomic manager is associated with
control of the autonomic element and actions like Monitoring, Analysis, Planning,
Execution using a Knowledge base. Managed elements on the other hand repre-
sent the hardware resources like storage, the CPU, etc.
The autonomic manager controls or influences the execution of the Managed
element and monitors its operations. Therefore it is a closed feedback loop
architecture.
The concept and functionality of the autonomic control architecture differs
from the “common” control architectures, and introduces the notions of self-
configuration, self-healing, self-optimization, and self-protection, working towards
building self-managing systems.
28 4 Biologically Inspired Robot Control Architecture
Fig. 4.5 Structure of autonomic element [KeC03].
Fig. 4.6 Generic Observer/Controller architecture.
4.3 ORCA (Organic Robot Control Architecture) 29
Within the Organic Computing initiative an Generic Observer/Controller archi-
tecture (Figure 4.6) has been firstly introduced in [RMB06], which incorporates
Observer and Controller units, where the Observer monitors the proper behavior
of Controller units and modifies their control in order to suit the predefined goals.
Observer/Controller architecture is a closed feedback loop architecture.
4.3 ORCA (Organic Robot Control Architecture)
ORCA development [BMM05] is a result of Organic Computing (Chapter 2.2)
research on developing hybrid robust robot control architecture that has self-x
properties and at the same time provides safe and reliable functioning.
Usually it is important for control architectures that exhibit self-x properties to
have the controlled emergence property. Namely, the system should be able to
learn and adapt its behavior, but at the same time not demonstrate some unwanted
behavior that exceeds some pre-defined constraints defined in the system’s core
specifications.
Fig. 4.7 ORCA – Organic Robot Control Architecture.
ORCA architecture is therefore built to satisfy these criteria, to provide reliable
robot function  control, to have modular architecture design, and to allow for
emergent properties of its constituent BCU and OCU units.
30 4 Biologically Inspired Robot Control Architecture
Basic Computing Units (BCUs) are basic software modules in the ORCA archi-
tecture, which may implement different functionalities related to robot control or
the robot’s hardware. These functionalities can be related to: sensor values acqui-
sition, sensor data fusion, sensor information pre-processing, or for example to
control a robot leg segment. Organic Computing Units (OCUs) are special type of
units in the ORCA architecture related to monitoring tasks for the correct behav-
iour of BCUs and also to control them to provide counteractions in case of anoma-
lies. ORCA architecture is presented in Figure 4.7.
ORCA is closely related to the research presented in this book, as a concept of
a robot control architecture.
4.4 Distributed ORCA Architecture for Hexapod Robot Control
ORCA has also been adapted to suit the practical experiments conducted on
experimental robots OSCAR-2 and OSCAR-X.
control
signals
feedback
signals
Fig. 4.8 Decentralized ORCA used in several robot experiments.
4.5 Cell Differentiation as Biological Inspiration for Enhanced ORCA 31
The idea behind practically implemented distributed organization of ORCA is
that each leg represented in ORCA consists of a number of BCU units which can
control some servo movements, and are related to some behaviors such as leg gait
generation, swing/stance leg movements, etc. There can be one OCU unit that is
responsible for monitoring the health status of the BCU units and providing coun-
teractions in case of detected anomalies. There is one OCU unit per leg, and there-
fore several of them in the whole architecture related to monitoring the behaviors
and health status of the leg units.
On the other hand, for anomaly detection purposes, a distributed ORCA can be
considered with one OCU per each BCU related to servo movement. Depending
on the application and task that the robot must realize, different types of distrib-
uted control architectures can be considered.
For example, the distributed ORCA architecture represented in Figure 4.8 has
been used in the experiments with emergent robot walking with distributed pres-
sure on its feet and for firefly inspired self-synchronization of walking gait of the
hexapod robot.
In the following chapter information is given on exploring new ideas about
ORCA architecture that would improve the robust and autonomic functioning of
this control architecture.
4.5 Cell Differentiation as Biological Inspiration for Enhanced
ORCA
An initial biologically inspired research was done on further enhancing the stan-
dard ORCA architecture, in order to provide means for achieving the self-x prop-
erties, like the self-organization of the modules by the robot control architecture.
A biological phenomenon called cell differentiation has been explored, more
specifically the self-organization that happens in this process. This study may aid
in improving the functionality of the ORCA control architecture.
First, a quick overview of the biological concept cell differentiation is appro-
priate before exploring the ideas of how these biological phenomena might be
beneficial for ORCA.
4.5.1 Overview of a Biological Concept – Cell Differentiation
By biological systems, the self-organization of the components seems to be intrin-
sically integrated. This can be seen when observing the cells in the organs and
tissues. Namely, the cells in one organism generally have some common charac-
teristics that can be observed in all the cells in the organism. For example, they all
carry the same DNA. But on the other hand, cells also do differ from each other. If
we observe the brain cells, muscles cells, liver cells, etc., they all differ in their
function or inner structure.
Research on types of cells called stem cells [BMT63] [SMT63] has given in-
sight into types of cells that are able to develop into different types of cells in the
32 4 Biologically Inspired Robot Control Architecture
body. Those stem cells are totipotent or pluripotent, and when dividing can poten-
tially (depending on the environment) transform and become any type of cell like
brain cell, muscle cell, liver cell, etc. This is done in the process of cell differentia-
tion [STE10], in which the cell due to various environmental conditions, inter-
signaling between the cells, or physical contact with neighboring cells can start to
develop into a specific type of cell.
4.5.2 The Enhanced “Stem” Type ORCA Architecture
If we have a look into ORCA and its OCU and BCU units and their pre-defined
functionality, we might consider these constitutive units as some sort of “already
differentiated cells” in the architecture. By ORCA structure mentioned earlier,
some BCU units can be related to motor control; some to control segments of the
robot; others can be related to provide sensor values read from the sensors, etc.
The OCUs can on the other hand be related to monitoring the individual BCU
units.
The idea behind utilizing the cell differentiation concept in this context would
be on introducing some “stem cell” like types of units in the architecture that can
“differentiate” into appropriate module types. This would be especially useful for
situations where other components like some servos and actuators (connected via
bus), should be dynamically installed into the robotic system without any need of
human operator intervention.
In an ordinary case such operation would require that the human operator iden-
tifies the type of module that has to be incorporated into the architecture and then
programs its interface so that the component can be suitably accessed. For exam-
ple, if it is an actuator, then it should be newly interfaced into the robotic system
and should receive commands for actions via its interface. If it is a sensor, then it
should provide data through its interface to the units that need it.
On the other hand, by the robot control architecture that has “stem” type of
units, it will not have them all preprogrammed and a fixed topology of the units’
interconnection. Some units at the start might be defined, for example some BCU
units controlling motors in the robot. But if any additional motors are added to the
robotic system, then by “fingerprint” of their functioning, they might appear to
have similarities with the already defined BCU units for the motor control. In that
case those “stem” type of units can be associated with BCU type interface for mo-
tor control.
Analogously this can be done also for any other extra sensors connected to the
robotic system. If a newly connected sensor device has a “fingerprint” that might
be similar in functionality to that of BCU units existing in the robotic system and
related to sensor signal acquisition, then such extra added sensor units will get
associated with BCU sensor type interface.
At first, all newly inserted hardware units are associated with some “stem” type
of units in the control architecture, and then they “differentiate” within the archi-
tecture to unit types in the control architecture that best describes their functional-
ity. When speaking of OCU units in the context of such enhanced “stem” ORCA
architecture, the “stem” type of OCU units might differentiate into other different
4.5 Cell Differentiation as Biological Inspiration for Enhanced ORCA 33
types of OCU units associated with monitoring of proper operation of associated
BCU units.
Such “stem” type of BCU and OCU units might be interesting to explore, since
upon replication and differentiation the newly generated and associated units will
perhaps express new intelligent functionality of the robotic system. On the other
hand these experiments with “stem” type of BCU and OCU units should be ap-
proached with great care, since there is also an open possibility that by such dif-
ferentiation the OCU units might expose an undesirable property or behavior of
the robotic system. There may be other techniques introduced that help to prevent
this from happening, like introducing special OCU monitor units that will “guide”
this process of differentiation to be compatible with the pre-defined functional
requirements of the robotic system.
The enhanced “stem” ORCA architecture model is presented in Figure 4.9.
Fig. 4.9 Enhanced “stem” type ORCA.
The whole mechanism of differentiation of the BCU and OCU units should be
explored, in the sense of defining a method for converting the “stem” type BCU or
OCU units. For example, they may be influenced by the neighboring units to be-
come a completely defined unit within the control architecture with a specific
function and interface.
34 4 Biologically Inspired Robot Control Architecture
One idea would be that the data oriented sequences originating from the OCU
and BCU units can be compared to the data streams originating from new “stem”
units. This might help the “stem” OCU or BCU units to be influenced to change
their type to that kind of BCU or OCU units that have similar data streams. This
would be analogue to the biological counterpart of cell differentiation, where the
different types of cells have or release slightly different types of molecules near
their vicinity, which differ from the molecules released in their surroundings by
other type of cells in the organism.
By the enhanced “stem” type ORCA organized in a bottom-up fashion, the dif-
ferentiation that occurs by the “stem” type of lower level BCU units and lower
level OCU monitoring units, is related more to the reactive behavior of the system.
In contrast, the higher level differentiated BCU and the higher level OCU units
(provided for their monitoring) are related to expressing the cognitive behavior of
the system. The rectangles show an example on how such elements would com-
municate with each other in such enhanced “stem” type ORCA.
B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 35–66.
springerlink.com © Springer-Verlag Berlin Heidelberg 2011
Chapter 5
Biologically Inspired Approaches for
Locomotion of a Hexapod Robot OSCAR
Different types of walking gait generation approaches have been considered for
walking by various multi-legged robots. Some of them are based on mathematical
formulations [PaH09] or inverse kinematic models [ShT07] trying to
mathematically model and describe the kinematics of the robot movements and
also the interaction of the robot with the environment. This may prove difficult
since completely modeling the robot and its interaction with the environment and
other environmental influences on proper working of robotic components is very
complex.
Another kind of approaches are biologically inspired. The biologically inspired
approach CPG (Central Pattern Generator) is based on walking patterns generated
by neural-networks [Mat87] seen by animals and insects. This is acquired for
producing walking movement by multi-legged walking robots [ITG09].
Another type of biologically inspired approaches can exhibit self-x properties
such as: self-organization or self-reconfiguration, similar to self-organization
properties seen in biological systems.
Before describing in detail the research done about such self-organizing
walking gait patterns based on emergence, several characteristics are given for
locomotion by insects and CPG based types of “common” walking gaits seen by
animals and insects.
5.1 Characteristics of Locomotion Seen by Insects and
Animals - Applied to Robotics Domain
5.1 Characteristics of Locomotion Seen by Insects and A nimals
Observations and research done on insects, arthropods and animals has provided
new insights on the locomotion seen in nature and on how living organisms
generate their movement patterns. These observations have been useful for
developing basic insect movements.
For example, research on the stick insect (Carausius morosus) [Cru76] walking
has given new information about the functionality of the leg segments and their
relation to protraction and retraction; elevation and depression; extension and
flexion.
36 5 Biologically Inspired Approaches for Locomotion of a Hexapod Robot OSCAR
There are two phases that are characteristic and essential for leg movement by
the insect: swing and stance phases. There are also two remarkable positions for
these movements – the posterior extreme position (PEP) – the position of the leg
on the ground when the leg is at the end of the power stroke; and the anterior
extreme position (AEP) – the position of the leg on the ground when the leg is at
the end of its return stroke.
In the swing phase, the leg is moving from the PEP to AEP, shown in Figure
5.1.1 – movement which precedes the movement of the leg over the ground. In
the stance phase, the leg is moving from AEP to PEP, which produces the thrust
that moves the insect over the ground. The swing and stance phases are
characterized with the length of their respective trajectories. The lengths of swing
and stance trajectories have direct influence on the speed with which the insect is
moving over the terrain. The longer the trajectories of swing and stance, the bigger
the distance travelled by the leg on the ground, and vice-versa, the shorter the
trajectories, the shorter the distance travelled on the ground. The length of swing
and stance phases by one leg in combination with swing and stance phases of
other legs may influence the insect to turn around its vertical axis.
As represented with circles in Figure 5.1, the leg of the insect can be considered
to be a 3 degrees of freedom (3 DOF) system. Namely, the leg consists of three
segments connected via joints, providing basic movements for protraction and
retraction; elevation and depression; extension and flexion.
Fig. 5.1 Swing and stance phases of an insect’s leg.
The same principle seen by leg structure and movement by insects has been
used for constructing and providing locomotion of the legs of multi-legged
walking robots. In Figure 5.2 (a), a 3 DOF structure is indicated by one of the
robot’s legs, which is similar to structure seen by the insect’s leg – the circles
represent the joints and the degrees of freedom. Servo “Alpha” is the nearest to the
body on each of the legs; servo “Beta” is next to it, down the leg; servo “Gamma”
is the last servo on the leg outward from the body. The swing and stance phases
and their trajectories by robot’s leg are presented in Figure 5.2 (b), which are
similar to swing and stance phases seen by an insect’s leg (Figure 5.1).
5.2 Central Pattern Generators (CPG) 37
(a) (b)
Fig. 5.2 (a) 3 DOF structure represented with circles on one of the robot’s legs; (b) Swing
and stance phases and their trajectories of the robot’s leg.
5.2 Central Pattern Generators (CPG)
Research on walking gait patterns has shown that neural networks called Central
Pattern Generators or CPG are located in the neural systems below the brain stem
[Gri81] in the spinal cord and are responsible for generating and modulating the
walking patterns and others specific to rhythmic motions.
Mathematical models for CPG are proposed in [Mat87] where the CPGs
consisted of two neurons, where each neuron receives an excitatory tonic input,
a mutual inhibition, and an external inhibitory input. At the end the output is
SELF INHIBITION
FROM OTHER
OSCILLATOR
TONIC
EXCITATION
FEEDBACK OUTPUT
MUTUAL
INHIBITION
+
-
Fig. 5.3 CPG model with two inhibiting neurons.
38 5 Biologically Inspired Approaches for Locomotion of a Hexapod Robot OSCAR
generated as a joint interaction of such inhibiting neurons. The mutual self-
inhibition is useful to induce a stable relaxation without any external input,
therefore an oscillation can be provided. Figure 5.3 represents the schematic of
such a model of two inhibiting neurons.
The neurons which mutually inhibit themselves can be related to flexion or
extension movements of the legs, which are important for walking. It has been
also shown that activities of CPGs can be modified by sensory feedback [CoB99]
and reflexes, which gives insight on how the rhythmic and reflex movements can
be coupled. Also Holk Cruse together with Friderich Pfeiffer did pioneer work
on a better understanding of biological control systems for 6 legged walking
machines (as well as Büschges, Berns, Ilg, Albiez,..).
5.2.1 Common Observed Gaits by Insects
Several walking gaits, which might be generated by an insect’s CPGs have been
observed: Wave gait, Ripple gait, and Tripod gait.
- Wave gait – is the slowest gait where one leg is in swing phase (in the air)
while the other legs are in the stance phase (on the ground). This gait is
characterized as the most stable one, since all the other legs are on the
ground and supporting the robot’s body. However, this is also the slowest
walking gait, since only one leg is in the air at a time, while the others are
on the ground.
- In Ripple gait – there are two independent gaits from both sides of the
body. The stance phase is usually double the swing phase and the opposite
legs are 180 degrees out of phase.
- The research on walking gaits of six legged insects, for example a
cockroach [SpM79], has indicated that the Tripod walking gait provides
the six-legged insect with the fastest speed over the ground. Tripod
walking gait is a gait by which at any moment of time three of the robot’s
legs are in the swing phase, while the other three legs are in the stance
phase. The three legs on the ground provide the insect with static and
dynamic stability while walking.
These gaits are represented in Figure 5.4 (a), (b), (c), going from the slowest one –
wave gait, then ripple gait, and lastly the fastest: tripod gait.
The black filled bars indicate the stance phase, while the non-filled ones
represent the legs during swing phases. On the symbolic represented insect in
Figure 5.4, the symbols L1, L2, L3 indicate the legs on the left side of the body
with their respective numbering and position on the body. The symbols R1, R2,
R3 indicate the legs on the right side of the body with their respective numbering
and position on the body. The symbols for left and right legs and their numberings
are located on the vertical axis, so it can be seen for which legs what type of leg
movement is taking place. The horizontal axis represents the time domain.
5.3 Experiments with Self-organizing Emergent Robot Walking Gait 39
Wave gait
(a)
Ripple gait
(b)
Tripod gait
(c)
Swing phase
Stance phase
R1
R2
R3
L1
L2
L3
Time
Fig. 5.4 Common gaits observed in insects (adapted from [FaC93]).
The presented insect walking gaits have been successfully applied for generating
robot walking for the domain of multi-leg walking robots patterns [IYS06]
[YAL06] [MDB07].
The fastest walking gait, tripod gait, has often been applied to robotic projects
to generate walking motion by the six legged walking robots. The tripod gait has
also been used within this research to provide the walking for the robot OSCAR-X
in some of the experiments which will be explained later in chapter 5.4.
5.3 Experiments with Self-organizing Emergent Robot Walking
Gait with Distributed Pressure on Robot’s Feet
5.3 Experiments wit h Self-organizing Emergent Robot Walking Gait
In the previous chapters 2.4 and 2.5, the terms about self-x properties and the
emergence were explained as being important for exploration and implementation
Other documents randomly have
different content
with minute decumbent hairs, less easily seen but still existing in the
secondary pair. In the Hymenoptera in general the wings are
covered with minute hairs or bristles; but in Tiphia, Scolia—with the
exception of S. Radula and affinities in which they are hairy—and
others, the wings are nearly naked; in Pompilus, Pepsis, c., the
hairs are infinitely numerous and very short; in the Sphecidæ,
Mutilla, c., they are more distinct, longer, and less numerous; in the
humble-bee (Bombus) and many others the apex of the wing is
darkened by a large number of more conspicuous hairs, each of
which seems to spring from a minute tubercle: as these tubercles
are in a part of the wing that is strengthened by few nervures, they
may probably be intended to supply their place, in giving firmness
and tension to this part. The wings of Diptera, under the present
head, may be viewed with regard to the hairs that are implanted in
the membrane of the wing, in its nervures, and in its margin. In the
first view, in Stratyomis and immediate affinities the wing is nearly
naked; but in Xylophagus, Beris, and the great majority of the Order,
the membrane of the wings is thickly planted with innumerable very
minute bristles, not to be seen but under a powerful lens, often
black, and seemingly crowning a little prominence, and giving the
wing an appearance of the finest net-work. As to the clothing of the
nervures, the costal, in Anthrax, Bombylius, c., is often remarkably
bristly at the base, with hairs intermixed; in Œstrus Ovis, in the inner
margin or edge of this nervure, is a single series of bristles, or rather
short spines, like so many black points; in Œ. Equi the whole costa is
covered with short decumbent hairs or bristles; in Musca pagana F.,
just at the apex of the costal areolet, that nervure is armed with a
spur or diverging bristle larger than the rest, which is also to be
found in many others of the Muscidæ, some of which have two and
others more of these spurs. The little moth-like midges (Psychoda
Latr., Hirtæa F.) at first appear to have the whole surface of their
wings covered with hairs; but upon a closer examination it will be
seen that they are planted in the nervures, from each of which they
diverge, so as under a lens to give it a very elegant
appearance[1956]. This fly has its wings beautifully fringed with fine
hairs, the third circumstance to be attended to under this head; in
the Tipulidans, and many others of this Order, the apex and
posterior margin are also finely fringed with short hairs. Some
Dipterous insects make a near approach to the Lepidoptera in the
covering of their wings: in the common gnat, when the wings are
not rubbed, the nervures are adorned by a double series of scales,
and the marginal fringe also consists of them[1957]; and in a
Georgian genus, which appears in some degree to connect Culex
with Anthrax c., there are scales scattered upon the membrane as
well as upon the nervures; besides, its antennæ[1958] and abdomen
are also covered with them.
The Order, the clothing of whose organs of flight excites the
admiration of the most incurious beholder, is that to which the
excursive butterfly belongs, the Lepidoptera. The gorgeous wings of
these universal favourites, as well as those of the hawk-moths and
moths, owe all their beauty, not to the substance of which they are
composed, but to an infinite number of little plumes or scales so
thickly planted in their upper and under surface, as in the great
majority entirely to conceal that substance. Whether these are really
most analogous to plumes or scales has been thought doubtful. De
Geer is inclined to think, from their terminating at their lower end in
little quills and other circumstances, that they resemble feathers as
much as scales[1959]; Reaumur on the contrary suspects that they
come nearer to scales[1960]. Their substance, approaching to
membrane, seems to make further for the former opinion, and their
shape and the indentations that often occur in their extremity,
furnish an additional argument for the latter. Their numbers are
infinite; Leeuwenhoek found more than 400,000 on the wings of the
silk-worm moth (Bombyx Mori)[1961]; and in those of some of the
larger moths and butterflies the number must greatly exceed this.
You will observe however that in many Lepidoptera the wings are
partially, and in some instances generally, transparent: thus in
Hesperia Proteus, a butterfly before noticed for the long tail that
distinguishes its secondary wings, there are many transparent spots;
in Attacus Atlas, one of the largest of moths, and its affinities, there
is as it were a window in each wing formed by a transparent
triangular space; in A. Polyphemus, Paphia, c., the pupil of the
ocellus is transparent, which in the former is divided by a nervure. In
several of the Heliconian butterflies, and in Zygæna F., c., the
greater part of both wings is transparent, with scales only upon their
nervures, round their margin, or forming certain bands or spots
upon them; in Parnassius Apollo, Mnemosyne, c., the scales are so
arranged as not wholly to cover the wings, which renders them
semidiaphanous; and in some (Nudaria) the wings are intirely
denuded. With regard to size, the scales vary often considerably in
different tribes; in Heliconia they appear to be more minute than in
the rest; and in Castnia they are the largest and coarsest; the
extremity of the wings of Lepidopterous insects in general is fringed
with longer scales than their surfaces, and even those of the last in
the same wing; sometimes vary in magnitude. The little seeming
tooth that projects from the middle of the posterior margin in the
upper wings of Notodonta, a subgenus of Bombyx L., is merely
produced by some longer diverging hairs. The shape and figure also
of scales are very various—some being long and slender; others
short and broad; some nearly round; others oval, ovate, or oblong;
others spathulate; others panduriform or parabolical; some again
almost square or rhomboidal; many triangular; some representing an
isosceles triangle, and others an equilateral one; lastly, some are
lanceolate and others linear; again, some have a very short pedicle
and others a very long one: with regard to their extremity; some are
intire, without projecting points or incisions, while others are
furnished with them: of these some terminate in a single long
mucro, others have several shorter ones; some are armed with
teeth, varying in number from two to thirteen in different
species[1962]. Many other forms might be enumerated, but these are
sufficient to give you a general notion of the infinite variety of this
part of the works of the Creator. I must next say a word or two upon
their arrangement on the wing. In most instances this is in
transverse lines, which sometimes vary a little from a rectilinear
course, and the extremity of the scales of one row reposes on the
base of those of the succeeding one, so that in this respect their
arrangement is like that of tiles in a roof: in some cases it is not so
regular: thus the minute scales on the wings of Parnassius Apollo,
and others with subdiaphanous wings, are arranged without order;
in Pieris and other Diurnal Lepidoptera, and many of the Crepuscular
and Nocturnal, there appears to be a double layer of scales on both
sides of the wing; the under layer usually consisting of white ones. If
you denude the wings of any butterfly, which you may easily do by
scraping it lightly on both sides with a penknife, you will be amused
to trace the lines in which the scales were planted, consisting of
innumerable minute dots: the lines of the under side, in some cases,
so cut those of the upper side, as by their intersection to form
lozenges. With regard to the position of the scales on the wing, they
usually lie flat, but sometimes their extremity is incurved: in the
beautiful Argynnis Vanillæ a very singular appearance of numerous
transverse ridges is produced by the extremity of those scales that
cover the longitudinal nervures of the primary wings, except at the
base, being recurved.
But though the general clothing of the wings of Lepidoptera consists
of these little scales, yet in some cases they are either replaced by
hairs or mixed with them. Thus, in the clear parts of the wings of
Heliconians, Attaci, c., short inconspicuous hairs are planted; in a
large number of the Orders the upper side of the Anal Area of the
secondary wings is hairy; in several Crepusculars (Sphinx Phœnix,
c.), where there is a double layer as before mentioned, the upper
one consists of dense hairs, except at the apex, and the lower one
of scales; and in most of them the scales of the primary wings are
piliform, and the secondary are covered by what approach very near
to real hairs; many of the Attaci are similarly circumstanced: the four
wings of A. Cytherea are also covered externally with hair.
7. Before I conclude this long diatribe on the organs of flight of
insects, I must not omit some notice of the infinite diversity of
colours with which their wings are often variegated and adorned by
the Creator, who loves to delight us by the beauty, as well as to
astonish and awe us by the immensity and grandeur of his works.
Though the wings in every Order exhibit instances of brilliant and
beautiful colouring, yet those of the Lepidoptera in this respect
infinitely excel them all, and to these, under this head, after noticing
a few in the less privileged Orders, my observations will be confined.
Although in the Coleoptera the wings are seldom distinguished by
their splendour; yet those of some Cetoniadæ, as Cetonia africana,
are extremely brilliant, and resemble those of many Xylocopæ in the
lovely violet hue that adorns them: amongst the Orthoptera some
Pterophyllæ, and in the Homopterous Hemiptera some Fulgoræ,
emulate the Lepidoptera in the ocelli that give a kind of life to these
organs[1963]; and a vast number of the destructive tribe of locusts
(Locusta Leach) are remarkable for the fine colours and gaiety of
their wings[1964]; in the Neuroptera numerous Libellulinæ emulate
the Heliconian butterflies by their maculation; and in the genus
Ascalaphus, which represents the Lepidoptera by its clubbed
antennæ[1965], many also have the resemblance increased by the
painting of their wings, so that some Entomologists have actually
considered some of them as belonging to that Order[1966]; the wings
of the Xylocopæ, before alluded to, sometimes add to the deep tints
of the violet—which also prevail in the wings of several Diptera—
towards their extremity the most brilliant metallic green or copper
varying,
As the site varies in the gazer's hand,
and even those wings that consist of clear colourless membrane are
often rendered extremely beautiful from the reflection of the
prismatic colours. I should undertake an endless task did I attempt
to specify all the modes of marking, clouding, and spotting, that
variegate a wing, and all the shades of colour that paint it, amongst
the Lepidopterous tribes; I shall therefore confine myself to a few of
the principal, especially those that distinguish particular tribes and
families. Of whole coloured wings—I know none that dazzle the eye
of the beholder so much as the upper surface of those of Morpho
Menelaus and Telemachus: Linné justly observes that there is
scarcely any thing in nature that for brightness and splendour can be
paralleled with this colour; it is a kind of rich ultramarine that vies
with the deepest and purest azure of the sky; and what must cause
a striking contrast in flight, the prone surface of the wings is as dull
and dark as the supine is brilliant, so that one can conceive this
animal to appear like a planet in full radiance, and under eclipse, as
its wings open and shut in the blaze of a tropical sun: another
butterfly, Papilio Ulysses, by its radiating cerulean disk, surrounded
on every side by a margin intensely black, gives the idea of light first
emerging from primeval obscurity; it was probably this idea of light
shining in darkness that induced Linné to give it the name of the
wisest of the Greeks in a dark and barbarous age. I know no insect
upon which the sight rests with such untired pleasure, as upon the
lovely butterfly that bears the name of the unhappy Trojan king (P.
Priamus); the contrast of the rich green and black of the velvet of its
wings with each other, and with the orange of its abdomen, is
beyond expression regal and magnificent. But peculiar beauties of
colour sometimes distinguish whole tribes as well as individuals.
What can be more lovely than that tribe of little butterflies that flit
around us every where in our summer rambles, which are called
blues, and which exhibit the various tints of the sky? Lycæna Adonis
of this tribe scarcely yields to any exotic butterfly in the celestial
purity of its azure wings: our native coppers also, Lycæna
dispar[1967], Virgaureæ, c., are remarkable for the fulgid colour of
these organs; in Argynnis the upper side of their wings is tawny,
spotted with black, while the under side of the secondary ones is
very often adorned by the appearance of silver spots. How this
remarkable effect of metallic lustre, so often reflected by spots in the
wings of butterflies, is produced, seems not to have occupied the
attention of Entomologists. M. Audebert is of opinion that the similar
lustre of the plumes of the humming birds (Trochilus) is owing to
their density, to the polish of their surface, and to the great number
of little minute concave mirrors which are observable on their little
beards[1968]. But these observations will not apply to the scales of
the wings of butterflies, which are always very thin and generally
very flat: in some instances, as in Morpho Menelaus, there appears
more than one very slight channel upon a scale; but this takes place
also in others that reflect no lustre. Their metallic hues must
therefore principally be occasioned by the high polish of their surface
and the richness of their tints. It is the purity of the white, in
conjunction with their shining surface, contrasted with the dull
opaque colour of the under side of the secondary wings, that causes
the spots that decorate those of the Fritillaries (Argynnis) to emulate
the lustre of silver. In Papilio the Trojans are distinguished by the
black wings with sanguine spots, and the Greeks by the same with
yellow spots; but these have proved in some instances only sexual
distinctions[1969]. In the Danai candidi L. the colour of the tribe may
be described as sacred to the day, since every shade, from white or
the palest yellow to full orange, is exhibited by them. The yellows
prevail also in those Noctuæ, the trivial names of which Linné made
to end in ago, as N. Fulvago, Citrago, c. I must not conclude this
part of my subject without noticing one of the most striking
ornaments of the wings of Lepidoptera, the many-coloured eyes
which decorate so large a number of them. Some few birds, as the
Peacock and Argus Pheasant, have been decked by their Creator
very conspicuously with this almost dazzling glory; but in the insects
just named it meets us every where. Some, as one of our most
beautiful butterflies, Vanessa Io[1970], have them both on the
primary and secondary wings; others, as Noctua Bubo[1971], only on
the primary; others again, as Smerinthus ocellata[1972], only on the
secondary: in some also they are on both sides of the wing, as in
Hipparchia Ægeria[1973], and in others only on the upper side, as in
Vanessa Io; in others again only on the under side, as in Morpho
Teucer[1974]: in some likewise they are very large, as in the
secondary wings of the same butterfly: and in others very small, as
in those in the wings of the blues (Lycæna). Once more, in some
they consist only of iris and pupil, as in Hipparchia Semele, and in
others of many concentric circles besides, as in Morpho Teucer, c.
v. Legs[1975]. We are next to consider those organs of motion affixed
to the trunk, by which insects transport themselves from one place
to another on the earth or in the water, and by which also they
perform various operations connected with their economy[1976]. In
treating of them we should consider their number; kind; substance;
articulation with the trunk; position; proportions; clothing;
composition; folding; and motions.
1. Number. Having before very fully explained to you the number
and kind of the legs of insects in their preparatory states[1977], I
shall now confine myself to the consideration of these organs in their
perfect or last state; beginning with their number. Insects, properly
so called, as I formerly observed[1978], in this state, including the
anterior pair or arms, have only six legs, none exceeding or falling
short of this number; but in several of the Diurnal Lepidoptera
(Vanessa, c.) the anterior pair are spurious, or at least not used as
legs, the tarsi having neither joints nor claws[1979]; this in some
cases is said to be only a sexual distinction[1980]. In Onitis,
Phanæus, and some other Scarabæidæ Mc
L., the arm has either
none or a spurious tarsus or manus[1981]; which in the first of these
genera is also a sexual character. From both these instances we see
that walking is only a secondary use of forelegs in the insect tribes.
Besides insects proper, a whole tribe of mites (Caris Latr., Leptus
Latr., Astoma Latr., Ocypete Leach) have only six legs; the rest, and
the Arachnida in general, have eight; in the Myriapods, Pollyxenus
has twelve pairs; Scutigera has fifteen; the terrestrial Glomerides (G.
Armadillo, c.) sixteen; and the oceanic (G. ovalis) twenty; the
oriental Scolopendræ Leach, twenty-one; Polydesmus has usually
about thirty pairs; Craspedosoma, fifty; Geophilus electricus at least
sixty; in Iulus terrestris there are more than seventy; in I. sabulosus
nearly one hundred; in I. fuscus, 124; and in I. maximus 134 pairs
or 268 single legs. But with respect to the Geophili, Iuli, c., it is to
be observed, that the number of pairs varies in different individuals;
and the circumstance that has been before mentioned[1982], that
these animals keep acquiring legs in their progress to the perfect
state, instead of losing them, renders it difficult to ascertain what is
the natural number of pairs in any species.
2. Kinds. Upon a former occasion I gave you a sufficiently full
account of the kinds of legs[1982], and I have also assigned my
reasons for giving a different denomination to the anterior legs
under certain circumstances[1983]; I shall not therefore enlarge
further upon this head.
3. Substance. The substance of the legs is generally regulated more
or less by that of the rest of the body, only in soft-bodied insects
they seem usually more firm and unbending. Each joint is a tube,
including the moving muscles, nerves, and air vessels.
4. Articulation with the Trunk. M. Cuvier has observed that the hip
(coxa), which is the joint that unites the leg with the body, rather
inosculates, in its acetabulum, than articulates in any precise
manner[1984]; but this observation, though true of a great many, will
not apply universally, for the legs of Orthopterous insects, and of
most of the subsequent Orders, are suspended rather than
inosculating. Even in many Coleoptera a difference is observable in
this respect. I have before mentioned that what are called the
puncta ordinaria, which distinguish the sides of the prothorax of
many Scarabæidæ and Geotrupidæ, form a base for an elevation of
the interior surface with which the extremity of the base of the
clavicle, which plunges deep into the breast, ginglymates[1985]; this
structure may also be found in other Lamellicorns, as the stag-beetle
(Lucanus) and Dynastes, that have not those excavations; in these
last it is an elevated ridge forming a segment of a circle with, it
should seem, a posterior channel, receiving a corresponding cavity
and protuberance of the clavicle. With regard to the mid-leg, in
Copris, the coxa is emboxed in a nearly longitudinal cavity of the
medipectus, and the coxa of the hind-leg anteriorly is suspended to
a transverse cavity of the postpectus, but posteriorly it is received by
a cavity of the first segment of the abdomen; so that it may be
regarded as suspended anteriorly, and inosculating posteriorly.
In some tribes of this Order, as the Weevils (Curculio L.) and
Capricorns (Cerambyx), the coxæ of the four anterior legs are
subglobose[1986] and extremely lubricous, and are received each by
a socket that fits it, and is equally lubricous. In the bottom of this
externally, and in the head of the coxa, is an orifice for the
transmission of muscles, nerves, and bronchiæ; but the coxa is
suspended by ligament in the socket. This structure approaches as
near the ball and socket as the nature of the insect skeleton will
permit; the high polish of the articulations acts the part of synovia,
and the motion is in some degree rotatory or versatile, whereas in
Copris, c., lately mentioned, it seems to be more limited, and is
probably, at least in the mid- and hind-legs, only in two directions; in
the middle pair, probably, from the coxæ being in a position parallel
with the breast, opposite to that of the hind pair. In Dytiscus L.,
Carabus L., and some other beetles, the coxæ, especially the
posterior pair, appear to be fixed and incapable of motion. In many
insects these coxæ seem to belong as much to the abdomen as to
the trunk. We have just seen this to be the case in Copris, c.; and
in the Lepidoptera, if the former be separated from the latter, the
legs will be detached with it.
4. Location. We are now to consider the location and position of the
legs, both in general and with respect to each other. And first, as has
been before stated, we may observe that, in the hexapods with
wings, the arms belong to the manitrunk, and are attached to the
antepectus on each side the prosternum; and the two pair of legs to
the alitrunk, the mid-legs being attached to the medipectus,
between the scapularia and mesosternum; and the hind-legs to the
postpectus, between the parapleura and the posternum; and further,
that the arms are opposed to the prothorax: the mid-legs to the
mesothorax and the primary organs of flight; and the hind-legs to
the metathorax and the secondary organs of flight; though in some
cases the wings appear to be behind the legs and in others before
them: thus, in Panorpa, the former are nearer the head than the
latter; but in the Libellulina the reverse of this takes place, the legs
being much nearer the head than the wings: in both cases, however,
the scapularia and parapleuræ run from the legs to the wings, but in
an oblique direction; and in Panorpa these pieces assume the
appearance of articulations of the legs. In most of the apterous
hexapods they appear to be attached laterally between the thorax
and the pectus[1987]; but in the flea (Pulex) they are ventral. In this
tribe the arms are usually stated to be inserted in the head[1988]:
but I once succeeded in separating the head of a flea from the
trunk, and these organs remained attached to the latter[1989]. As to
the Octopods and Arachnidæ, in the mites (Acarus L.) they are
lateral, and in their analogues, the spiders (Aranea L.), they emerge
between the thorax and the breast, which last they nearly surround;
in the Phalangidæ the bases of the coxæ approach near to each
other, being separated only by a narrow sternum; in their
antagonists, Chelifer and Scorpio, they apply to each other, the
anterior ones acting as maxillæ. In the myriapods the legs of the
Chilopoda Latr., and some Chilognatha, as Glomeris, are inserted
laterally, a single pair in a segment; but in Iulus L. their attachment
is ventral, the coxæ seem to spring from a common base, and there
are two pair to each segment[1990], except the three first, which
bear each a single pair.
I shall next consider how the legs are located with respect to each
other. To render this clear to you I shall represent each of the
variations, which amount in all to twelve in the hexapods that have
fallen under my notice, by six dots.
1. In this arrangement the legs are all planted near to each
other, there being little or no interval between the pairs, and
between the legs of each pair. It is exemplified in the Lepidoptera,
Blatta, and many Diptera.
2. Similar to the preceding, but the anterior pair are distant from
the two posterior; exemplified in the bees (Apis) and most
Hymenoptera; Chironomus; Scutellera; Pachysoma K.[1991]
3. Like the last, but the posterior pair is distant from the two
anterior. Examples: Silpha, Necrophorus, Telephorus, c.
4. Similar to the last, but the legs of the posterior pair are more
distant from each other than the four anterior. Ex. Curculio L.
5. The legs of each pair near each other, but the pairs distant. Ex.
Gibbium.
6. Both the legs of each pair and the pairs distant. Ex. Blaps, c.
7. Anterior pair distant from the two posterior, and the legs of
the middle pair rather more distant from each other than those of
the other pairs. Ex. Scarabæus Mc
L.
8. Like the preceding, only the legs of the middle pair are at a
much greater distance from each other. Ex. Copris Mc
L.
9. Legs of the two posterior pairs distant. Ex. Hister, Scaphidium.
10. Like the preceding, but the posterior legs more distant than
those of the middle pair. Ex. Lygæus.
11. Like the last, but the legs of the anterior pair also distant. Ex.
Velia.
12. The arms distant, intermediate legs more distant, posterior
legs close together. Ex. Byrrhus L.
5. Proportions. In general the legs of some insects are
disproportionally long and slender, as in Phalangium Opilio and some
species of Gonyleptes[1992]: those of others are disproportionally
short, as in Elater, c. With regard to their relative proportions, the
most general rule is, in Hexapods, that the anterior pair shall be the
shortest and most slender, and the posterior the longest and
thickest; but there are many exceptions: thus, in Macropus
longimanus, Clytra longimana, c., in the male the arms are the
longest; again, a thing that very rarely occurs, in the same sex of
Podalirius retusa the intermediate legs are the longest[1993]; but in
Rhina barbirostris and many weevils they are the shortest: in
Saperda hirtipes Oliv.[1994] the hind-legs are disproportionally long:
with regard to thickness, they are in general extremely slender in
Cicindela, and in the Scarabæidæ very thick. In Goliathus Cacicus
the arms are more robust than the four legs[1995]; in Gyrinus the
latter are more dilated than the former; in many Rutelidæ, and
particularly in the celebrated Kanguroo beetle (Scarabæus Macropus
Franc.) the hind-legs are much the thickest; in a new genus of
weevils from Brazil (Plectropus K.), the intermediate pair are more
slender than either the arms or the posterior pair.
6. Clothing. The hairs on the legs of insects, though at first sight
they may seem unimportant, in many cases are of great use to
them, both in their ordinary avocations and motions: but as most of
these were sufficiently noticed when I treated of the sexes of
insects[1996], I shall not here repeat my observations, but confine
myself to cases not then adverted to. Some insects have all their
legs very hairy, as many spiders, the diamond beetle (Entimus
imperialis), or at least a species very near it and common in
Brazil[1997], c.: in others they are nearly naked, as in the stag-
beetle. In the Crepuscular Lepidoptera (Sphinx L.) and some of the
Nocturnal ones (Bombyx L.) the thighs are much more hairy than the
rest of the legs: and in Lucanus, Geotrupes, and many other
Lamellicorns, c., the anterior ones have a yellow or golden spot at
their base, composed of decumbent hairs, which prevent them from
suffering by the violent friction to which they are exposed in
burrowing. In most Petalocerous beetles the tibiæ are set with
scattered bristles, and sometimes the thighs. The Tiger beetles
(Cicindela) are similarly circumstanced: but the bristles, which are
white, are generally arranged in rows. In Dytiscus, Hydrophilus, c.,
the four posterior tarsi; and in Notonecta the posterior pair, and also
the tibiæ—are fringed on each side with a dense series of hairs,
which structure assists them in swimming[1998]. The tarsi, especially
the anterior pair, in a certain family of Lamia F. (L. papulosa, c.
[1999]), are similarly fringed, only the hairs curl inwards; and the
hand in Sphex and Ammophila, but not in Pelopæus and Chlorion, is
fringed externally with long bristles.
7. Composition. With regard to their composition, both arms and legs
generally consist of five pieces, which Entomologists have
denominated—the coxa or hip—the trochanter—the femur or thigh—
the tibia or shank—and the tarsus or foot. Where the structure and
use of the fore-leg is different from that of the four hind-legs, I
propose calling these pieces by names corresponding with those
which anatomists have appropriated to the arm in the higher
vertebrate animals: thus, as you will see in the table, I call the whole
fore-leg the brachium or arm; and the coxa becomes the clavicula or
collar-bone; the trochanter, the scapula or shoulder-blade; the
femur, the humerus or shoulder; the tibia, the cubitus or arm; the
tarsus, the manus or hand. But let me not lead you to suppose that
the pieces, either in the arms or legs of insects, which are there
named after certain others in vertebrate animals, precisely
correspond with them—by no means—since that is a very doubtful
point; and some of them, as the trochanter, clearly do not. Many
gentlemen skilled in anatomy, as I have before observed[2000], have
thought that what is regarded as the coxa in insects really
represents the femur: but there are considerable difficulties in the
way of this supposition, several of which I then stated. I shall not
however enter further into the subject, and take the above names;
since this application of them is so general and so well understood,
except with regard to the fore-leg, under certain circumstances, as I
find them. I shall now consider them in the order in which I have
named them.
a. Coxa or Clavicula[2001]. The coxa is the joint that connects the leg
with the trunk of the insect. With regard to their shape, the most
general form of the four anterior is more or less that of a truncated
cone: in the Staphylinidæ, however, they tend to a pyramidal or
four-sided figure; as do the whole six in the Trichoptera: in numbers
of the weevils and capricorns they are subglobose; in the
Lamellicorns they are mostly oblong, and not prominent: the
posterior pair in the Coleoptera are generally flat and placed in a
transverse position, and more or less oblong and quadrangular: in
Elater, c., they are cuneiform: in Haliplus Latr. they are dilated, and
cover the thigh[2002]: in Buprestis, Copris, c., they have a cavity
that partly receives it: the corresponding part, the clavicle, in the
arm of Gryllotalpa, is very large and remarkable; viewed underneath
it is triangular, and trifid where the trochanter articulates with it: in
that of Megachile Willughbiella the clavicle is armed with a
spine[2003]. As to their proportions, the most general law seems to
be, that the anterior pair shall be the shortest and smallest, and the
posterior the longest and largest. In some instances, as in Buprestis,
the two anterior pair are nearly equal; in others (Mantis, Eurhinus
K.), the anterior are the longest, in the former as long as the thigh,
and the four posterior the shortest: in the Trichoptera, Lepidoptera,
c., all are nearly equal; in Mantis the two posterior, and in
Phengodes the intermediate pair are the largest; but in Necrophorus
they are the smallest:—though almost universally without
articulations, in Galeodes the clavicle consists of two and the coxa of
three[2004].
b. Trochanter or Scapula[2005]. This is the second joint of the leg:
and if the coxa is regarded as the analogue of the thigh in vertebrate
animals, this should seem to represent the patella or rotula, vulgarly
called the knee-pan. Latreille and Dr. Virey consider this articulation
as merely a joint of the coxa[2006]; but if closely examined,
especially in Coleopterous insects, you will find it so fixed to the
thigh as scarcely to have separate motion from it, and in many cases
it seems to be merely its fulcrum; but I am not aware that any
instance occurs in which it has not motion separate from that of the
former joint.
As to its articulation with the coxa,—in the Coleoptera it appears to
be of a mixed kind; for it inosculates in that joint, is suspended by
ligament to its orifice, and its protuberances are received by
corresponding cavities in it; and its cavities receive protuberances,
which belongs to a ginglymous articulation. I have observed two
variations in this Order, in one of which the motion of the thigh and
trochanter is only in two directions, and in the other it is nearly
versatile or rotatory. The Lamellicorns afford an example of the first,
and the Rhyncophorous beetles or weevils of the second. If you
extract from the coxa the thigh with the trochanter of the larger
species of Dynastes Mc
L., you will find that the head of the latter is
divided into two obtuse incurving lobes or condyles: that on the
inner side being the smallest and shortest, and constricted just
below its apex: and that under this is a shallow or glenoid cavity,
terminating posteriorly in a lubricous flat curvilinear ridge. If you
next examine the trochanter in articulation with the coxa, you will
perceive that the head of the former inosculates in it, that the lower
condyle is received by a sinus of the coxa, which also has a lubricous
very shallow cavity corresponding with the ridge, in which it turns;
and in the head of the coxa, on the lower side, is an external
condyle, which is received by a sinus common to both, of the head
of the thigh and of the exterior side of the trochanter[2007], in which
it likewise turns: this last condyle has also an internal protuberance,
which appears to ginglymate with a cavity of the trochanter: from
this structure the leg is limited chiefly to a motion up and down upon
two pivots, or to fold and extend itself. You will find an articulation
very near this, but on a smaller scale, in the stag-beetle. In the
other kind of articulation, which admits of freer motion, the head of
the trochanter is prolonged, and the process terminates in a short
interior condyle, which appears to work in a corresponding cavity of
the interior of the coxa; and the base of the process is encompassed
by a ridge with a cavity behind it, which is received by another of
the lower part of that piece, and admits a corresponding ridge—a
structure that allows a rotatory motion. In the hind-legs of this tribe
the motion is chiefly limited to folding and extending; in Carabus,
c., also the head of the trochanter is nearly hemispherical, and the
articulation approaches ball and socket. In most of the other Orders,
the Hymenoptera excepted, there is little or no inosculation, the
trochanter being simply suspended by ligament to the coxa as well
as to the thigh; its connection with the latter is similar in Coleoptera;
but in Cicindela, c., it inosculates in it. The part we are considering
varies in its position with respect to the thigh: in the hind-legs of
Carabus, c., it forms a lateral fulcrum on the inner side of that part,
and does not intervene between its base and the coxa; the muscles
from the latter entering the former, not at the bottom of the base,
but at its side: but in the four anterior legs it forms their base, as it
does in all the legs in Apion, and in all the Orders except the
Coleoptera, cutting them entirely off from contact with the coxa: in
the Lamellicorns they cut off part of the base obliquely, but so as to
permit their coming in contact with the condyle of the coxa, as
before mentioned. In the Ichneumonidæ and some other
Hymenoptera the trochanter appears to consist of two joints
particularly visible in the posterior legs[2008].
As to size in general,—the part in question is smaller than the coxa;
but in Notonecta it is larger, and in the dog-tick (Ixodes Ricinus)
longer than that joint. It exhibits few variations in its shape or
appendages worthy of particular notice. In general, in the Coleoptera
it is triangular or trigonal; but in Carabus L., in the hind-leg it is
oblong or rather kidney-shaped; in that of Necrophorus[2009] it
terminates in one or two teeth or spines, varying in length in the
different species: in the other Orders it is not remarkable in this
respect.
c. Femur or Humerus[2010]. The femur or thigh is the third, and
usually the largest and most conspicuous joint of the leg. In the
hypothesis before alluded to[2011] it is considered as the analogue of
the tibia of vertebrate animals. With regard to the articulation of this
part with the trochanter, it has been sufficiently explained under that
head, and that with the tibia I shall treat of when I come to that
joint. As to the size of the thighs, and their relative proportions to
each other and to the remaining joints of the leg, the most general
law is, that the anterior pair shall be the shortest and smallest, and
the posterior the longest and largest. With respect to the remaining
articulations, most commonly the thigh is longer and larger than the
tibia, and the tibia than the tarsus. But there are numerous
exceptions to both these rules. With respect to the first, we may
begin by observing that the increase of the magnitude of the thigh,
from the anterior to the posterior pair, is usually gradual: but in
many jumping insects, and likewise many that do not jump, the
posterior pair are suddenly and disproportionally thicker than the
rest[2012]. Again, in many insects the anterior pair are the longest
and thickest, as in Macropus longimanus, Bibio, Nabis, c.: in
others, the intermediate exceed the rest in magnitude, as in Onitis
Aygulus, cupreus; Sicus flavipes, c.; in many Lamellicorns all the
thighs are incrassated and nearly equal in size: but in some, as
Ryssonotus nebulosus Mc
L.[2013], the intermediate pair are rather
smaller than the rest. With respect to the second rule—in some, as
in the male of Macropus longimanus, the anterior tibia, though more
slender, is longer than the thigh; in Hololepta maxillosa it is longer
and more dilated; in Lamia marmorata, or one related to it from
Brazil, the intermediate pair are longer; in Ateuchus gibbus and
others of that tribe the posterior thighs are smaller than the tibiæ:
and, to mention no more; in Callichroma latipes the posterior tibia is
wider than the part last named. Again, the tarsi are as long as either
tibia or thigh in many of the larger Dynastidæ, as Megasoma
Actæon, c.; longer than either in Melolontha subspinosa F.; and in
Tiphia, Scolia and affinities, often as long, or longer than both
together.
As to shape,—the thigh, especially in the fore-leg, varies
considerably: most generally it is flat, linear, and a little thicker
where it is united to the tibia, on the outer side convex, and concave
next the body; but in many it is gradually thicker from the base to
the apex: in some Cerambyces (C. thoracicus) it is clavate; in others
of this genus and Molorchus they may be called capitate; in
Pterostichus they are rather lanceolate; in Onitis Sphinx the humerus
is triangular, and the intermediate thigh rhomboidal; in Bruchus
Bactris it is bent like a bow; and in some Brazilian Halticæ it is nearly
semicircular. The humerus in Phasma is attenuated at the base; in
Empusa gongyloides it is at first ovato-lanceolate, and terminates
below in a kind of footstalk[2014]; in Phasma flabelliforme it is
dolabriform[2015]; in Mantis often semioval or semielliptical, and
thickest at the inner edge, which affords space for two rows of
spines with which it is planted. In Phyllium siccifolium all the thighs
are furnished on both sides with a foliaceous appendage nearly from
base to apex[2016]: in a species of Empusa (E. macroptera), the four
posterior ones are so distinguished only on their posterior side[2017]:
others of this last genus, as E. gongyloides, have an alary
appendage on both sides at the apex of these thighs[2018]; and
another family, as E. pauperata, have only one on the posterior
side[2019]. The thighs of no insect are more remarkable for their
elegant shape,—tapering gradually from the base to the apex, where
they swell again into a kind of knee,—than the posterior ones of the
locusts (Locusta Leach); each side of these thighs is strengthened
with three longitudinal nearly parallel ridges, and the upper and
under sides are adorned by a double series, in some coalescing as
they approach the tibia, of oblique quadrangular elevations
resembling scales[2020].
I shall next say a few words upon the spines and other processes
which arm the thigh. Those moveable ones of Mantis which help to
form a fearful instrument of destruction, have just been mentioned,
and similar ones, but less conspicuous, arm the intermediate thighs
of Sicus flavipes: other appendages of this kind are for a less
destructive purpose—to keep the tibia when folded in its place. This
seems to be the use of the serratures and spine that arm the thigh
of Bruchus Bactris, or the Hymenopterous genera Leucospis, Chalcis,
c.; in Onitis Aygulus a short filiform horn arms the humerus, and a
longer crooked one that of many species of Scaurus[2021]. In many
Stenocori the thighs terminate in two spines, and in Gonyleptes K.
the posterior ones are armed internally with very strong ones; with
which, as the legs converge at their knee[2022], they may probably
detain their prey. The knee-pan (Gonytheca) of the thigh, or the
cavity at its end, which receives the head of the tibia, is very
conspicuous in the weevils; but in no insects more than in
Locusta[2023], in which tribe it deserves your particular attention.
d. Tibia or Cubitus[2024]. The tibia or shank is the fourth joint of the
leg, which according to the hypothesis lately alluded to is the
analogue, in the anterior leg of the carpus or carpal bones, and in
the four posterior ones of the tarsus or tarsal bones of vertebrate
animals. This may be called the most conspicuous of the
articulations of the leg; for though it is generally more slender and
often shorter than the thigh, it falls more under the eye of the
observer, that joint being more or less concealed by the body: it
consists in general of a single joint; but in the Araneidæ and
Phalangidæ it has an accessory one, often incrassated at its base,
which I have named the Epicnemis[2025].
With respect to the articulation of the tibia with the thigh—we may
observe that in general it is by means of three processes or
condyles, two lateral and one intermediate, of the head of the
former joint[2026]: the lateral ones are usually received by a cavity or
sinus of the gonytheca of the thigh[2027]; and upon these the tibia
turns, with a semirotatory motion up and down as upon a pair of
pivots: at the same time the mola or head of the latter joint, which
has often a flexure so as to form an elbow with the rest of it,
inosculates in the gonytheca, and is also suspended by ligament to
the orifice through which the muscles, nerves, and bronchiæ are
transmitted: so that in fact the articulation, strictly speaking, belongs
exclusively to none of the kinds observable in vertebrate animals,
but partakes of several, and may properly be denominated a mixed
articulation,—a term applicable in numerous instances also to the
other articulations of the legs of insects. In the different Orders
some variations in this respect take place,—I will notice some of the
most remarkable. In no Coleopterous insects is the structure more
distinctly visible than in the larger Lamellicorns. In Copris
bucephalus, for instance, if you divide the thigh longitudinally, you
will find on each side, at the head, that it is furnished with a nearly
hemispherical protuberance, perforated in the centre for the
transmission of muscles, and surrounded externally by a ridge,
leaving a semicircular cavity between them[2028]: if you next
examine the tibia, after having extracted it, you will find on each
side, at the base, a cavity corresponding with the protuberance of
the thigh which it receives, having likewise a central orifice, and
surrounded by a semicircular ridge corresponding with the cavity in
the thigh in which it acts: below this ridge another cavity, forming a
small segment of a circle, receives the ridge of the thigh[2029]. You
will observe that the ridge of the tibia represents the lateral condyle
lately noticed: in the Dynastidæ this is more prominent, and often
forms a smaller segment of a circle. In these also the protuberance
of the thigh is more minute, and its ridge is received by a cavity of
the tibia nearly semicircular[2030]; in Geotrupes Latr. the articulation
is not very different, though on a reduced scale; in Calandra
Palmarum the lateral condyles of the tibiæ are flatter and
broader[2031]; and the articulation not being quite so complex, this
joint is kept steady by an intermediate process observable in the
gonytheca[2032]. From the above description it appears that the
dislocation of the tibia is effectually prevented in the Lamellicorns by
the protuberance and ridge of the thigh working in their
corresponding cavities, while the condyle of that part turns with a
rotatory motion in the cavity of the thigh. In the Orthoptera Order
the tibia is suspended by a ligament, in the gonytheca the lateral
condyles, which are very prominent, working in a sinus of that
part[2033]. The subsequent Orders exhibit no very striking variations
from these types of articulation, I shall therefore not detain you
longer upon this head.
With regard to the proportions and magnitude of the joint we are
considering,—the most general law is, that the anterior pair should
be shorter and more slender than the intermediate; and the
intermediate than the posterior; and that all the tibiæ should be
shorter and more slender than the thighs, and longer and thicker
than the tarsi. Various exceptions, however, to this rule in all these
cases might be produced; but I shall only observe that in all those
insects in which the fore-legs are calculated for digging or seizing
their prey, as in the Petalocerous beetles, the Gryllotalpa, Mantis,
c., this joint of the leg is usually much enlarged and more
conspicuous than the others.
As to its figure and shape—most commonly the tibia grows thicker
from the base to the apex, as in the majority of Coleoptera,
Hymenoptera, c.; in the Orthoptera, Neuroptera, c., it is generally
equally thick every where. Another peculiarity relating to this head
observable in it, is its tendency to a trigonal figure: this, however,
though very general, is not universal;—thus, in some Orthoptera, as
Pterophylla K., its horizontal section is quadrangular; in others, as
Locusta Leach and many other insects, it is nearly a circle; in some
scorpions it is almost a hexagon. The superficial shape also of this
joint in numerous instances is more or less triangular, but it
sometimes recedes from this form:—thus, in Callichroma latipes it is
a segment of a circle; in some Empides it is clavate; in Onitis Sphinx,
dolabriform; in the Orthoptera, Neuroptera, c., it is usually linear;
in some Lygæi it is angular[2034]: but the most remarkable tibiæ in
this respect are those of such species of this last genus as have the
posterior ones winged or foliaceous, so that they resemble the leaf
of some plant—the tibia being the rachis, and the wing (which in
some species is veined) representing the leaf itself. This structure is
exemplified in Lygæus compressipes, phyllopus, foliaceus, c.[2035]
Under this head I must say a few words upon the flexure of this
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com

More Related Content

PDF
Advanced Robotics For Medical Rehabilitation Current State Of The Art And Rec...
PDF
Wheeled Mobile Robotics From Fundamentals Towards Autonomous Systems 1st Edit...
PDF
Agile Autonomy Learning Highspeed Visionbased Flight Antonio Loquercio
PDF
Recent Advances In Mobile Robotics A Topalov
PDF
Multirobot Systems Trends And Development T Yasuda K Ohkura
PDF
Experimental Robotics The 14th International Symposium On Experimental Roboti...
PDF
Bioinspiration And Robotics Walking And Climbing Robots Optimized Maki K Habib
PDF
Bioinspired Sensing Actuation And Control In Underwater Soft Robotic Systems ...
Advanced Robotics For Medical Rehabilitation Current State Of The Art And Rec...
Wheeled Mobile Robotics From Fundamentals Towards Autonomous Systems 1st Edit...
Agile Autonomy Learning Highspeed Visionbased Flight Antonio Loquercio
Recent Advances In Mobile Robotics A Topalov
Multirobot Systems Trends And Development T Yasuda K Ohkura
Experimental Robotics The 14th International Symposium On Experimental Roboti...
Bioinspiration And Robotics Walking And Climbing Robots Optimized Maki K Habib
Bioinspired Sensing Actuation And Control In Underwater Soft Robotic Systems ...

Similar to Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth (20)

DOCX
Emerging trends in robotics using neural network
PDF
Recent Advances In Multirobot Systems Aleksandar Lazinica
PDF
New Developments And Advances In Robot Control 1st Ed Nabil Derbel
PDF
Computer And Information Science Roger Lee
PDF
Computer And Information Science Roger Lee
PDF
Exoskeletons In Rehabilitation Robotics Tremor Suppression 1st Edition Eduard...
PDF
Robotics For Cell Manipulation And Characterization 1st Changsheng Dai
PDF
MODEL PREDICTIVE CONTROL BASED JUMPINGOF ROBOTIC LEG ON A PARTICULAR HEIGHT U...
PDF
Multiagent Robotic Systems 1st Edition Jiming Liu
PDF
Multiagent Robotic Systems 1st Edition Jiming Liu
PDF
Prototyping Of Robotic Systems Applications Of Design And Implementation 1st ...
PDF
Advances In Control Of Articulated And Mobile Robots Bruno Siciliano Et Al
PDF
Soft Robotics In Rehabilitation Amir Jafari Nafiseh Ebrahimi
PDF
Multiagent Robotic Systems 1st Edition Jiming Liu
PDF
Walking Biological And Technological Aspects 1st Edition Friedrich Pfeiffer
PDF
Humanrobot Interaction Control Analysis And Design Dan Zhang
PDF
Active sensors for local planning in mobile robotics 1st Edition Penelope Pro...
PDF
Active sensors for local planning in mobile robotics 1st Edition Penelope Pro...
PDF
Cabledriven Parallel Robots Proceedings Of The Second International Conferenc...
PDF
Nonfunctional Properties In Service Oriented Architecture Requirements Models...
Emerging trends in robotics using neural network
Recent Advances In Multirobot Systems Aleksandar Lazinica
New Developments And Advances In Robot Control 1st Ed Nabil Derbel
Computer And Information Science Roger Lee
Computer And Information Science Roger Lee
Exoskeletons In Rehabilitation Robotics Tremor Suppression 1st Edition Eduard...
Robotics For Cell Manipulation And Characterization 1st Changsheng Dai
MODEL PREDICTIVE CONTROL BASED JUMPINGOF ROBOTIC LEG ON A PARTICULAR HEIGHT U...
Multiagent Robotic Systems 1st Edition Jiming Liu
Multiagent Robotic Systems 1st Edition Jiming Liu
Prototyping Of Robotic Systems Applications Of Design And Implementation 1st ...
Advances In Control Of Articulated And Mobile Robots Bruno Siciliano Et Al
Soft Robotics In Rehabilitation Amir Jafari Nafiseh Ebrahimi
Multiagent Robotic Systems 1st Edition Jiming Liu
Walking Biological And Technological Aspects 1st Edition Friedrich Pfeiffer
Humanrobot Interaction Control Analysis And Design Dan Zhang
Active sensors for local planning in mobile robotics 1st Edition Penelope Pro...
Active sensors for local planning in mobile robotics 1st Edition Penelope Pro...
Cabledriven Parallel Robots Proceedings Of The Second International Conferenc...
Nonfunctional Properties In Service Oriented Architecture Requirements Models...
Ad

Recently uploaded (20)

PPTX
IMMUNITY IMMUNITY refers to protection against infection, and the immune syst...
PDF
Chapter 2 Heredity, Prenatal Development, and Birth.pdf
PDF
Supply Chain Operations Speaking Notes -ICLT Program
PDF
Insiders guide to clinical Medicine.pdf
PPTX
Cell Structure & Organelles in detailed.
PDF
TR - Agricultural Crops Production NC III.pdf
PPTX
master seminar digital applications in india
PDF
Complications of Minimal Access Surgery at WLH
PDF
ANTIBIOTICS.pptx.pdf………………… xxxxxxxxxxxxx
PDF
Classroom Observation Tools for Teachers
PPTX
Lesson notes of climatology university.
PPTX
1st Inaugural Professorial Lecture held on 19th February 2020 (Governance and...
PPTX
Pharmacology of Heart Failure /Pharmacotherapy of CHF
PPTX
GDM (1) (1).pptx small presentation for students
PDF
grade 11-chemistry_fetena_net_5883.pdf teacher guide for all student
PDF
FourierSeries-QuestionsWithAnswers(Part-A).pdf
PDF
Saundersa Comprehensive Review for the NCLEX-RN Examination.pdf
PDF
Basic Mud Logging Guide for educational purpose
PDF
Pre independence Education in Inndia.pdf
PDF
The Lost Whites of Pakistan by Jahanzaib Mughal.pdf
IMMUNITY IMMUNITY refers to protection against infection, and the immune syst...
Chapter 2 Heredity, Prenatal Development, and Birth.pdf
Supply Chain Operations Speaking Notes -ICLT Program
Insiders guide to clinical Medicine.pdf
Cell Structure & Organelles in detailed.
TR - Agricultural Crops Production NC III.pdf
master seminar digital applications in india
Complications of Minimal Access Surgery at WLH
ANTIBIOTICS.pptx.pdf………………… xxxxxxxxxxxxx
Classroom Observation Tools for Teachers
Lesson notes of climatology university.
1st Inaugural Professorial Lecture held on 19th February 2020 (Governance and...
Pharmacology of Heart Failure /Pharmacotherapy of CHF
GDM (1) (1).pptx small presentation for students
grade 11-chemistry_fetena_net_5883.pdf teacher guide for all student
FourierSeries-QuestionsWithAnswers(Part-A).pdf
Saundersa Comprehensive Review for the NCLEX-RN Examination.pdf
Basic Mud Logging Guide for educational purpose
Pre independence Education in Inndia.pdf
The Lost Whites of Pakistan by Jahanzaib Mughal.pdf
Ad

Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth

  • 1. Biologically Inspired Approaches For Locomotion Anomaly Detection And Reconfiguration For Walking Robots 1st Edition Bojan Jakimovski Auth download https://ebookbell.com/product/biologically-inspired-approaches- for-locomotion-anomaly-detection-and-reconfiguration-for-walking- robots-1st-edition-bojan-jakimovski-auth-2451800 Explore and download more ebooks at ebookbell.com
  • 2. Here are some recommended products that we believe you will be interested in. You can click the link to download. Biologically Inspired Approaches To Advanced Information Technology First International Workshop Bioadit 2004 Lausanne Switzerland January 2930 2004 Revised Selected Papers 1st Edition Jacqueline Signorini https://ebookbell.com/product/biologically-inspired-approaches-to- advanced-information-technology-first-international-workshop- bioadit-2004-lausanne-switzerland-january-2930-2004-revised-selected- papers-1st-edition-jacqueline-signorini-2330358 Biologically Inspired Approaches To Advanced Information Technology Second International Workshop Bioadit 2006 Osaka Japan January 2627 2006 1st Edition James C Liao Auth https://ebookbell.com/product/biologically-inspired-approaches-to- advanced-information-technology-second-international-workshop- bioadit-2006-osaka-japan-january-2627-2006-1st-edition-james-c-liao- auth-4140854 Computational Intelligence In Multifeature Visual Pattern Recognition Hand Posture And Face Recognition Using Biologically Inspired Approaches 1st Edition Pramod Kumar Pisharady https://ebookbell.com/product/computational-intelligence-in- multifeature-visual-pattern-recognition-hand-posture-and-face- recognition-using-biologically-inspired-approaches-1st-edition-pramod- kumar-pisharady-4698730 Biologically Inspired Control Of Humanoid Robot Arms Robust And Adaptive Approaches 1st Edition Adam Spiers https://ebookbell.com/product/biologically-inspired-control-of- humanoid-robot-arms-robust-and-adaptive-approaches-1st-edition-adam- spiers-5483834
  • 3. Evolutionary Optimization Algorithms Biologicallyinspired And Populationbased Approaches To Computer Intelligence Dan Simon https://ebookbell.com/product/evolutionary-optimization-algorithms- biologicallyinspired-and-populationbased-approaches-to-computer- intelligence-dan-simon-36520926 Biologically Inspired Cognitive Architectures 2010 Proceedings Of The First Annual Meeting Of The Bica Society 1st Edition A V Samsonovich K R Jhannsdttir A Chella https://ebookbell.com/product/biologically-inspired-cognitive- architectures-2010-proceedings-of-the-first-annual-meeting-of-the- bica-society-1st-edition-a-v-samsonovich-k-r-jhannsdttir-a- chella-51692978 Biologically Inspired Cooperative Computing Ifip 19th World Computer Congress Tc 10 1st Ifip International Conference On Biologically Inspired Cooperative In Information And Communication Technology 1st Edition Yi Pan https://ebookbell.com/product/biologically-inspired-cooperative- computing-ifip-19th-world-computer-congress-tc-10-1st-ifip- international-conference-on-biologically-inspired-cooperative-in- information-and-communication-technology-1st-edition-yi-pan-2141834 Biologically Inspired Networking And Sensing Algorithms And Architectures Pietro Li https://ebookbell.com/product/biologically-inspired-networking-and- sensing-algorithms-and-architectures-pietro-li-2490906 Biologicallyinspired Optimisation Methods Parallel Algorithms Systems And Applications 1st Edition Kalyanmoy Deb Auth https://ebookbell.com/product/biologicallyinspired-optimisation- methods-parallel-algorithms-systems-and-applications-1st-edition- kalyanmoy-deb-auth-2543422
  • 6. Cognitive Systems Monographs Volume 14 Editors: Rüdiger Dillmann · Yoshihiko Nakamura · Stefan Schaal · David Vernon
  • 8. Bojan Jakimovski Biologically Inspired Approaches for Locomotion, Anomaly Detection and Reconfiguration for Walking Robots ABC
  • 9. Rüdiger Dillmann, University of Karlsruhe, Faculty of Informatics, Institute of Anthropomatics, Humanoids and Intelligence Systems Laboratories, Kaiserstr. 12, 76131 Karlsruhe, Germany Yoshihiko Nakamura, Tokyo University Fac. Engineering, Dept. Mechano-Informatics, 7-3-1 Hongo, Bukyo-ku Tokyo, 113-8656, Japan Stefan Schaal, University of Southern California, Department Computer Science, Computational Learn- ing & Motor Control Lab., Los Angeles, CA 90089-2905, USA David Vernon, Khalifa University Department of Computer Engineering, PO Box 573, Sharjah, United Arab Emirates Author Dr.-Ing. Bojan Jakimovski Bionics4Robotics Postfach 900609, 81506 München, Germany E-mail: contact@bionics4robotics.com ISBN 978-3-642-22504-8 e-ISBN 978-3-642-22505-5 DOI 10.1007/978-3-642-22505-5 Cognitive Systems Monographs ISSN 1867-4925 Library of Congress Control Number: 2011934501 c 2011 Springer-Verlag Berlin Heidelberg This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable for prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Typeset Cover Design: Scientific Publishing Services Pvt. Ltd., Chennai, India. Printed in acid-free paper 5 4 3 2 1 0 springer.com
  • 10. Abstract The increasing presence of mobile robots in our everyday lives introduces the re- quirements for their intelligent and autonomous features. Therefore the next gen- eration of mobile robots should be more self-capable, in respect to: increasing of their functionality in unforeseen situations, decreasing of the human involvement in their everyday operations and their maintenance; being robust; fault tolerant and reliable in their operation. Although mobile robotic systems have been a topic of research for decades and aside the technology improvements nowadays, the subject on how to program and making them more autonomous in their operations is still an open field for research. The classical formal methodologies which have been dominating the robotics segment for long time start to prove that they are perhaps not adequate to cope with the increasing complexity of the robotic systems. Many different research directions have been considered on how to overcome these problems. Applying bio-inspired, organic approaches in robotics domain is one of the methodologies that are considered that would help on making the robots more autonomous and self-capable, i.e. having properties such as: self-reconfiguration, self-adaptation, self-optimization, etc. In this book several novel biologically inspired approaches for walking robots (multi-legged and humanoid) domain are introduced and elaborated. They are related to self-organized and self-stabilized robot walking, anomaly detection within robot systems using self-adaptation, and mitigating the faulty robot conditions by self-reconfiguration of a multi-legged walking robot. The ap- proaches presented have been practically evaluated in various test scenarios, the results from the experiments are discussed in details and their practical usefulness is validated.
  • 12. Contents 1 Introduction ................................................................................................... 1 2 Biologically Inspired Computing and Self-x Properties............................. 5 2.1 Bionics..................................................................................................... 5 2.2 Organic Computing ................................................................................. 6 2.3 Autonomic Computing ............................................................................ 6 2.4 Self-x Properties ...................................................................................... 6 2.5 Emergence ............................................................................................... 7 3 Joint Leg Walking and Hybrid Robot Demonstrators............................... 9 3.1 Introduction ............................................................................................. 9 3.2 Hexapod Robots....................................................................................... 9 3.2.1 State of the Art – Hexapod Robots.............................................. 11 3.2.2 Hexapod Robot Demonstrator – OSCAR (Organic Self Configuring and Adapting Robot) ............................................... 11 3.3 Humanoid Robots .................................................................................. 17 3.4 State of the Art Humanoid Robots......................................................... 17 3.5 Humanoid Robot Demonstrator - S2-HuRo (Self Stabilizing Humanoid Robot) ....................................................... 19 4 Biologically Inspired Robot Control Architecture.................................... 23 4.1 Overview on “Standard” Types of Robot Control Architectures........... 24 4.1.1 Reactive and Subsumption and Behavior Based Control Architecture ................................................................................ 24 4.1.2 Deliberative Control Architecture............................................... 25 4.1.3 Hybrid Control Architecture....................................................... 26 4.2 Overview on Autonomic Control Architecture...................................... 27 4.3 ORCA (Organic Robot Control Architecture)....................................... 29 4.4 Distributed ORCA Architecture for Hexapod Robot Control................ 30 4.5 Cell Differentiation as Biological Inspiration for Enhanced ORCA...... 31 4.5.1 Overview of a Biological Concept – Cell Differentiation........... 31 4.5.2 The Enhanced “Stem” Type ORCA Architecture....................... 32 5 Biologically Inspired Approaches for Locomotion of a Hexapod Robot OSCAR.............................................................................................. 35 5.1 Characteristics of Locomotion Seen by Insects and Animals - Applied to Robotics Domain................................................. 35 5.2 Central Pattern Generators (CPG) ......................................................... 37
  • 13. VIII Contents 5.2.1 Common Observed Gaits by Insects........................................... 38 5.3 Experiments with Self-organizing Emergent Robot Walking Gait with Distributed Pressure on Robot’s Feet..................................... 39 5.4 Firefly Inspired Synchronization of a Robot’s Walking Gait ................ 45 5.4.1 Firefly Coupled Oscillators Principle ......................................... 46 5.4.2 Concept for Robot Walking Gait Self-synchronization by Using Firefly Synchronization.................................................... 47 5.5 Implementation of Firefly Inspired Self-synchronization into the Robot Control Architecture ................................................................... 55 5.6 Experiments Done with Firefly Inspired Self-synchronization and Results from Experiments...................................................................... 56 5.6.1 Experiment about Self-synchronization by Prolongation of the Robot’s Swing and Stance Phases .............................................. 57 5.6.2 Experiment about Self-synchronization by Shortening of the Robot’s Swing and Stance Phases .............................................. 60 5.6.3 Experiment about Self-synchronization by Combined Prolongation and Shortening of the Robot’s Swing and Stance Phases ....................................................................... 62 5.6.4 Discussion on Future Possible Improvements of Firefly Inspired Self-synchronization Approach .................................... 66 5.6.5 Summary about the Firefly Inspired Self-synchronization Approach..................................................................................... 66 6 Biologically Inspired Approach for Optimizing the Walking Gait of a Humanoid Robot.......................................................................................... 67 6.1 Approaches for Walking Gait Generation by Humanoid Robots........... 67 6.2 Symbiosis as a Biologically Inspired Approach for Self-stabilization of Humanoid Robot Walking Gait............................. 69 6.3 SelSta Approach in Detail...................................................................... 71 6.3.1 S2-HuRo Humanoid Robot Platform and Sensors Used............. 71 6.3.2 Control of the Robot S2-HuRo.................................................... 72 6.3.3 Main Parts of SelSta Approach – SymbScore Value and Genetic Algorithm ....................................................................... 74 6.3.4 Fuzzy Logic Computation of SymbScore Value......................... 78 6.3.5 Genetic Algorithm Details for the SelSta Approach ................... 84 6.3.6 Preparation for Experiments........................................................ 85 6.4 Experiments Done with the SelSta Approach........................................ 88 6.4.1 Experiments on a Soft Green Carpet........................................... 89 6.4.2 Experiments on a Medium Soft Orange Carpet .......................... 97 6.4.3 Experiments on a Hard Green Carpet ....................................... 106 6.4.4 Experiments on a Hard Linoleum Surface................................ 115 6.5 Summary for Experiments Done with the SelSta Approach................ 124 7 Biologically Inspired Approaches for Anomaly Detection within a Robotic System........................................................................................... 127 7.1 Overview on Approaches for Fault / Anomaly Detection by Robotic Systems .................................................................................. 127
  • 14. Contents IX 7.2 Overview of Artificial Immune System (AIS) Concept ...................... 128 7.3 Artificial Immune System Based - Robot Anomaly Detection Engine (RADE) Approach................................................................... 131 7.3.1 Core Functionality of RADE Approach.................................... 134 7.4 Experiments Done with AIS Inspired RADE and Results from Experiments......................................................................................... 135 7.4.1 Test-Bed Setup for RADE Approach........................................ 135 7.4.2 Self and Non-self Rule Sets by RADE ..................................... 136 7.4.3 Results from Experiments Done with the RADE Approach..... 138 7.4.4 3D representation of Run-Time Dynamics by RADE Anomaly Detection Surface....................................................... 147 7.4.5 Summary about AIS Based Anomaly Detection Approach - RADE .................................................................... 150 8 Approach for Robot Self-reconfiguration after Anomaly Detection within a Walking Robot System Based on Biological Inspiration - Swarm Intelligence............................................................. 151 8.1 Overview on Swarm Intelligence – Flocking Behavior and Boids...... 152 8.2 S.I.R.R. – Swarm Intelligence Based Approach for Robot Reconfiguration ................................................................................... 155 8.2.1 Simulation of S.I.R.R Based Hexapod Robot Reconfiguration ........................................................................ 158 8.3 Results from Robot Reconfiguration Experiments Done with S.I.R.R. Approach on the Hexapod Robot OSCAR-2 ......................... 161 8.4 Results from Real Robot Reconfiguration Experiments Done with S.I.R.R. Approach and Leg Amputations on the Robot OSCAR-X .... 162 8.4.1 Ground Contacts of Robot Legs for Normal Walking and for Walking with Leg Amputations and Robot Self-reconfiguration................................................................. 164 8.4.2 Tracking of the Robot’s Heading While the Robot Is Performing Self-reconfiguration with Leg Amputations .......... 166 8.5 Summary for the S.I.R.R. - Biologically Inspired Robot Reconfiguration Approach................................................................... 173 9 Conclusion and Outlook............................................................................ 175 10 References................................................................................................. 179 A Appendix.................................................................................................... 189 A.1 Test Bed for Tracking the Robot OSCAR-X during the Experiments ....................................................................................... 189 List of figures .................................................................................................... 191 Keywords........................................................................................................... 199 Glossary............................................................................................................. 201
  • 16. B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 1–3. springerlink.com © Springer-Verlag Berlin Heidelberg 2011 Chapter 1 Introduction Robotic systems nowadays are getting increasingly complex in their design and implementation. In order to fulfill the proposed requirements, systems often con- sist of many software and hardware units, realizing various functionalities and cooperating together. Declaring them as autonomous means that they should have the ability to dynamically adjust and execute their tasks without human intervention. Additionally, they should be reliable and also tolerant to various system malfunctions. Future robotic systems should be able to demonstrate self-x properties such as self-organization, self-reconfiguration, self-healing and the like. Having these kinds of properties, robotic systems would be able to demonstrate their autonomic property, and this will also aid in shortening their development and maintenance time. However, the complexity of the classical approaches for robot system modeling has introduced a need for developing and applying new concepts and methodolo- gies towards creating self-capable, more robust, and dependable systems. In order to achieve this, engineers have used different biologically and organically inspired approaches. For example, some of the algorithms for motor control and walking gait pattern generations for the domain of joint leg walking robots have been inspired and developed on observations seen within animals and functioning of the neural circuitry. In that context, research has been done and is presented in this book on introducing novel biologically inspired approaches with their practical applications for domain of self-organizing, self-optimizing, and self-reconfiguring walking robots. Joint leg walking robots and simple walking robots use their legs for their movement over a terrain. Depending on how many legs they have, they can belong to one of many categories, from two legged humanoid robots up to many legged robots having three, four, six, or eight legs, where each leg can be built out of several joints/segments. Walking robots can have many joints and therefore many degrees of freedom (DOF). However, having many degrees of freedom introduces difficulties in developing appropriate methods for controlling such complex walking robots, monitoring their health status, making them fault- tolerant, etc. The research presented in this book tries to give an answer on some of the open questions for fault-tolerant robotics domain. This includes self-organized,
  • 17. 2 1 Introduction self-stabilized robot walking and self-adapting methods for failure detection and self-reconfiguration after a failure has been detected within the robotic system, so the robotic system can still continue with its mission despite the faulty conditions within itself. Research presented in this book includes: - biologically inspired robot control architecture; - self-organizing hexapod robot walking; - biologically inspired self-stabilizing humanoid robot walking; - biologically inspired self-adapting approach for anomaly detection by a robot; - biologically inspired approach for self-reconfiguration by a hexapod robot. The structure of the thesis is organized as follows: - In the 2nd chapter, a general overview is given on biologically inspired com- puting and self-x properties including short introduction to terms related to: bionics, organic computing, autonomic computing, self-x properties, and emergence; - The 3rd chapter is about joint leg walking and hybrid robots, with a small review about state of the art of hexapod robots. In this chapter detailed de- scriptions about the hexapod robot demonstrators OSCAR-2, OSCAR-3, and OSCAR-X are given. State of the art humanoid robots are further in- troduced and also a detailed description of the S2-HuRo humanoid robot demonstrator is given. - The 4th chapter describes notions about a biologically inspired robot con- trol architecture. An overview is given on commonly used robot control architectures such as: reactive, subsumption, deliberative, and hybrid con- trol. After the introduction on commonly used robot control architectures, an introduction is given on self-organizing robot architectures such as: autonomic control architecture, Organic Robot Control Architecture (ORCA), and their characteristics. The distributed ORCA which is related to the research experiments is explained in detail and new ideas are also given about an enhanced “stem” based ORCA architecture. - The 5th chapter first gives an introduction on locomotion seen by insects and animals and Central Pattern Generator (CPG) for walking pattern gen- eration. Then a concept for self-organizing emergent robot walking gait with distributed pressure on robot’s feet is explained. Further explained in this chapter is the firefly-inspired synchronization of a robot’s walking gait. Included are the experiments done on prolongation and shortening of the robot’s swing and stance phases using firefly-inspired synchronization. - In the 6th chapter a biologically inspired approach for optimizing the walk- ing gait of a humanoid robot is explained. Symbiosis as a biologically inspired approach for self-stabilization of humanoid robot walking gait is elaborated and details about the SelSta approach and its main parts are
  • 18. Introduction 3 discussed. The chapter concludes with the results of experiments done on the SelSta approach and the usefulness of the SelSta approach. - The 7th chapter is about biologically inspired approaches for failure detec- tion within a robotic system. First it gives an overview on the approaches for robot fault / anomaly detection, followed by an introduction on Artificial Immune Systems. Then it introduces the Artificial Immune System (AIS) based Robot Anomaly Detection Engine (RADE) approach. After RADE is explained in more detail, results from the experiments done with the AIS- inspired RADE approach are presented and discussed. - The 8th chapter introduces an approach for robot self-reconfiguration of a hexapod robot system based on an biological inspiration - swarm intelli- gence. An overview of swarm intelligence, flocking behavior, and boids is given first. Then S.I.R.R., a Swarm intelligence based approach for robot reconfiguration, is introduced and explained in detail. After the introduc- tion of the S.I.R.R. approach, results from simulation of S.I.R.R.-based hexapod robot reconfiguration are presented. The results from real robot reconfiguration experiments done with the S.I.R.R. approach on the hexa- pod robot OSCAR-2 are then presented for validation of the simulation experiment’s results. Chapter 8 ends with the results from real robot recon- figuration experiments done with the S.I.R.R. approach, presentation of leg amputations on robot OSCAR-X, and an explanation of practical useful- ness of S.I.R.R. - The 9th chapter gives a conclusion on the research presented in this book and the importance of the biologically-inspired approaches introduced for the walking and general robotics domain.
  • 20. B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 5–7. springerlink.com © Springer-Verlag Berlin Heidelberg 2011 Chapter 2 Biologically Inspired Computing and Self-x Properties 2.1 Bionics Bionics is the application of biological methods and systems found in nature to the study and design of engineering systems and modern technology [Bio10]. The term Bionics (from biology and electronics) is sometimes interchangeably used for a Biomimetics and Biomimicry (from bios = life, and mimesis = to imi- tate). Bionics is related to applying ideas seen in nature for solving scientific, technical, or engineering problems. Biomimetics is therefore an interdisciplinary field where scientists from different scientific fields like Biology, Physics, Chem- istry, and Engineering work together towards developing various solutions based on observations of processes seen in nature. All these techniques are based on solving new problems from solutions to previous problems found in nature. For example, such biologically inspired concepts can be related to various organizational building principles seen within bacteria, flora, fauna, etc. There are many innovations and products developed that can be mentioned as examples for practical usefulness of biologically inspired concepts: self- assembling glass inspired by sea sponges; bacterial control inspired by red algae; solar cells inspired by leaves; friction-free fans inspired by nautilus; building ma- terial from CO2 inspired by mollusks; self-cleaning surfaces inspired by lotus plant [Nat10], etc. Besides the natural assembling techniques which can be useful for building novel materials with new and perhaps superior properties than that of the current existing materials, the social and organizational principles in nature like feeding, mating, foraging, and swarming are also found to be useful for engineering domain. Bionics is more related to implementing the approach found in nature as an idea instead of imitating the biologically structure behind it, which is more closely representing the Biomimetics. Nowadays there is an increasing trend of applying and adopting biologically in- spired approaches for the general domain of computer science as well as for the domain of robotics.
  • 21. 6 2 Biologically Inspired Computing and Self-x Properties More and more of the approaches implemented in the field of computer engi- neering are related to artificial intelligence and optimization, such as artificial neu- ral networks, genetic algorithms, and ant optimization algorithms. 2.2 Organic Computing Organic computing initiative [Mül04] [Sch05], is related to development of tech- nical systems that act autonomously and dynamically adapt to the environment. Those systems must exhibit lifelike properties and function in independent way. That is why they are called “organic” or “organic computing systems”. At the same time they should be also robust, safe, and trustworthy. Therefore, organic computing is a type of biologically inspired computing that attempts to develop approaches for technical systems exhibiting self-x properties such as self-organization, self-configuration, self-optimization, self-healing, and self-explaining. The research presented in this book is directly associated with Organic Com- puting and developing self-x biologically inspired approaches that would enable the robotic systems to act in more independent and autonomous ways. 2.3 Autonomic Computing The Autonomic computing initiative was proposed in the IBM [IBM01] manifesto and states the need for development of autonomic IT systems that would over- come the ever growing complexity of current IT systems. The main requirement for such autonomic systems would be that they are self-manageable and also capable of providing reliable services and minimizing the human administrator intervention and thus minimizing the probability of human errors. Autonomic systems are therefore systems that can manage themselves without human intervention. They must be capable of incessant autonomous work given only high-level objectives from the administrators [KeC03]. In order to achieve the autonomic system's property, such systems must have the self-x properties self-reconfiguration, self-organization, and self-healing. 2.4 Self-x Properties Self-x properties are closely related to biological processes found in nature, namely the processes in nature that can be often seen as self-organizing, self- optimizing, and self-healing processes. Such self-x properties have been proven useful when translated to scientific domains: techniques for new materials development, engineering, new approaches for the IT industry, etc. Approaches mentioned in this book are mainly related to the development of various algorithms for joint leg walking robots domain that enable the robots to
  • 22. 2.5 Emergence 7 exhibit the so-called self-x properties for various circumstances such as robot self- reconfiguration, robot walking gait self-optimization, and robot self-healing. General definitions for the terms Self-Configuration, Self-Optimization, Self- Healing, and Self-Protection can be found by Autonomic Computing [KeC03] and also can be used interchangeably for other technical domains, as well as for the robotics domain. Self-Configuration: Automated configuration of components and systems follows high-level policies. The rest of the system adjusts automatically and seamlessly. Self-Optimization: Components and systems continually seek opportunities to improve their own performance and efficiency. Self-Healing: The system automatically detects, diagnoses, and repairs localized software and hardware problems. Self-Protection: The system uses early warning to anticipate and prevent system- wide failures. 2.5 Emergence Emergence is one of the phenomena often observed in nature. Emergence can be defined as “the arising of novel and coherent structures, patterns, and properties during the process of self-organization in complex systems [Gol10]. The emer- gence can sometimes be summarized as: Whole is more the than sum of its parts. This means that in systems exhibiting emergence, the behavior or the property of the whole system cannot be deducted from the properties of individual compo- nents composing that system. Such a definition is closer to the view of “strong emergence”. On the other side, “weak emergence” is related to the emergence that is traceable, i.e. the emergent property can be reduced to the property of individual components. For emergence is often said to be a “bottom-up” process. There are many examples of emergent processes that can be seen in nature or in biological systems. For example: the sand dunes, water waves, swarming schools of fish, flocking of birds, slime molds, ant colonies’ self-sustainability, etc. In complex systems where safety is not a critical issue (since the completely safe system's behavior cannot be guaranteed), emergence is sometimes used to lower the effort of developing the needed system's functionality. Emergence is also popular for domain of robotic systems. Examples include: emergence of gait patterns for robot walking [AWY99], emergence of communica- tion within multi robot systems [Lip07], emergent behaviors of autonomous robots [AnD90], etc. The practical usefulness of the emergence concepts is explained in further chapters, applied to the research on the hexapod robot OSCAR, more precisely for its walking gait generation.
  • 24. B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 9–21. springerlink.com © Springer-Verlag Berlin Heidelberg 2011 Chapter 3 Joint Leg Walking and Hybrid Robot Demonstrators 3.1 Introduction Design of two and multi-legged robots like four legged, six legged, and eight legged robots shows a practical usefulness of Bionics for the domain of robotics. Depending on the number of legs, such robotic platforms can be inspired from body constitution, walking mechanics, and behavior of humans, animals (four legged), insects (six legged), or spiders (eight legged). There is also another kind of robotic designs depicted as hybrid robots, which have a mixture of concepts seen in nature and artificial designs. These robots can be often found having tracks, other special leg designs, wheeled-legged robot designs and the like. In this chapter several joint leg walking and hybrid robots are discussed. The “joint leg” term is related to robots that have legs built out of servos, representing their joints. The term “hybrid” found by hybrid robots is related to robots that have both legs built out of servo joints and wheels attached to their legs. Robots demonstrators presented in this chapter have been used as demonstra- tors for biologically inspired approaches and algorithms researched and elaborated in this book. These robots can be categorized into three types: humanoid robots, hexapod-robots, and hybrid wheeled-legged robots. 3.2 Hexapod Robots Hexapod robots belong to the group of joint leg walking robots having six legs where the legs are consisting of multiple servo joints. The legs of the robot are usually symmetrically distributed in two different groups spatially located on the two opposite sides of the robot's body. The design of hexapod robots is often in- spired by locomotion systems seen in insects like cockroaches, stick insects, and the like. In comparison with the four legged walking robots or quadruped robots, hexapod robots have intrinsically more redundancy due to the higher number of the legs and thus can be theoretically more flexible over uneven terrain. Hexapod
  • 25. 10 3 Joint Leg Walking and Hybrid Robot Demonstrators robots differ from robots that have “native” spider-like biomimetic design having eight legs distributed on the two sides of the robot’s body. Although the eight legged robots may have higher degree of redundancy and perhaps provide better agility for the robot over rugged terrain, they also need more energy for their func- tioning, which in turn affects the size and mobility of the robot. (a) (b) (c) (d) (e) Fig. 3.1 Hexapod robots: (a) “iSprawl”; (b) “RHex”; (c) “DLR Crawler”; (d) “RiSE”; (e) “AMOS-WD06”.
  • 26. 3.2 Hexapod Robots 11 3.2.1 State of the Art – Hexapod Robots There have been many types of hexapod robots that have been used for demon- stration purposes in the research on biologically inspired locomotion. Some of the current (Feb, 2010) state of the art hexapod robots include: “iSprawl” [KCC06], “RHex” [AMK01], “DLR Crawler” [GWH09], “RiSE” robot [SGF06], “AMOS- WD06” [STW10] (Figure 3.1). However, the mentioned hexapod robots all differ in the technology that they use for their locomotion. For example, their leg design differs from one robot to the other and the moving concept of the joints by the legs also differs. Movement of the legs for the “iSprawl” robot is periodic, generated by push- pull actions using flexible cables and servo motors [KCC06]. For the “RHex” robot, the movement of the legs is related only to the rotary motion of the legs [AMK01]. The “DLR Crawler” design was based on the “DLR-HAND II” [BFH03], therefore the joint based fingers of the “DLR-HAND II” are adapted to serve as legs for the “DLR Crawler” [GWH09]. For the “RiSE” robot, the legs are moved by two electrical actuators per leg, with biologically inspired adhesive structures located on the feet, which enable the “RiSE” robot to climb on vertical wall surfaces, trees, etc [SCM06]. The “AMOS-WD06” robot implements legs consisting of three servos each, resulting in eighteen servos for locomotion of the hexapod robot [STW10]. By the presented state-of-the-art robots there are improvements that can be seen in comparison with some older hexapod robot designs, however less has been done on introducing fault tolerant mechanisms within the robots itself, which will give the robots the robustness to be functional also in situations when they experience some malfunctions within their components, by means of reconfigur- ing the body and leg postures. And this is the key difference in comparison to the robot demonstrators OSCAR, in particular OSCAR-X, which were developed at Institute für Technische Infor- matik, University Lübeck and described in the following sub-chapter. 3.2.2 Hexapod Robot Demonstrator – OSCAR (Organic Self Configuring and Adapting Robot) OSCAR (Organic Self Configuring and Adapting Robot) is a six legged walking robot, used as a demonstrator for testing some of the newly-developed biologically inspired approaches and algorithms presented in this book. OSCAR represents the series of built hexapod robot demonstrators used in the interdisciplinary research, where the legs are distributed spatially in a circle on the robot's body. Most of the robots in OSCAR series have integrated an on-board embedded system, sensors (ultrasonic, infrared, acceleration, and inclination sen- sors), and actuators (analog and digital servos). Such a typical spatial distribution of the robot’s legs in a circle is represented in (Figure 3.2).
  • 27. 12 3 Joint Leg Walking and Hybrid Robot Demonstrators Fig. 3.2 Spatial distribution in circle of legs by OSCAR series of robots. The OSCAR series consists of the following robots: OSCAR-1, OSCAR-2, OSCAR-3, and OSCAR-X, which are described below. The first two robots, OSCAR-1 and OSCAR-2, were mostly based on the Lynxmotion® robot kit - “AH3-R (18 Servo Walker)” with 18 degrees of freedom (DOF), Hitec HS-645 Servos and aluminum based leg design [Lyn06]. 3.2.2.1 Hexapod Robot Demonstrator – OSCAR - 1 Hexapod robot OSCAR-1, built in year 2006 is the first in the series of OSCAR robots. Its hardware is based on the “AH3-R (18 Servo Walker)” robot kit with six legs distributed spatially in a circle and additional on-board electronics such as “JControl” (a Java based embedded system for robot control), servo controller SD- 21, Hitec analog servos HS-645, binary contact sensors on the robot's feet, and NiMH batteries (Figure 3.3). Each leg by the robot is made up of three servos. There are also integrated ultrasonic sensors on three of the robot’s legs. Fig. 3.3 Hexapod robot OSCAR-1
  • 28. 3.2 Hexapod Robots 13 3.2.2.2 Hexapod Robot Demonstrator – OSCAR - 2 OSCAR-2 (Figure 3.4) is the second in the series of OSCAR robots, similarly built as OSCAR-1. The OSCAR-2 in comparison to the OSCAR-1, has the following modifications: - pressure sensors (Figure 3.5); - 18 modified HiTec HS-645 servos (Figure 3.6); The modified servos provide feedback for the level of servo current, so the torque can be monitored while the robot is walking. The modification is clearly visible due to the number of wires that come out from the servos, namely that the wires are directly connected to the potentiometer output reading pins inside of the servos. (a) (b) Fig. 3.4 Hexapod robot OSCAR-2. (a) Experimental robot OSCAR-2 setup - from above; (b) Robot OSCAR-2 in movement. Fig. 3.5 Pressure sensors type FSR-400. The most right one in the figure is used by OSCAR 2. Another difference to OSCAR-1 is that OSCAR-2 has pressure sensors on its feet (Figure 3.5) instead of the binary contact sensors, so a variable pressure on the robot’s feet can be sensed. Furthermore, OSCAR 2 has an accelerometer sensor used to sense the accelera- tion and inclination of the robot.
  • 29. 14 3 Joint Leg Walking and Hybrid Robot Demonstrators By experiments with OSCAR-2, National Instruments hardware [Nat06] and software was used for acquisition and pre-processing of the signals (currents from servos, feet pressure, inclination values), their graphical representation, and data logging. Fig. 3.6 Modified HiTec HS-645 servo with wires for current and position feedback Other important characteristic for robot OSCAR-2 is that in order to simulate a faulty situation of the robot’s legs by anomaly detection experiments, there have been modifications introduced to some of the robot’s legs for those particular ex- periments. Such experimental leg modification is shown in (Figure 3.7). Fig. 3.7 Modification by leg of robot OSCAR-2, in order to allow simulated leg failure The presented modification allows the robot’s leg to intentionally malfunction (the inserted pins drop off) after some time of robot walking. 3.2.2.3 Hexapod Robot Demonstrator – OSCAR - 3 OSCAR-3 is similar to OSCAR-1 and OSCAR-2 in its principal construction. The difference to OSCAR-1 and OSCAR-2 is that the 18 modified HiTec HS-645 ser- vos have additional internal electronic printed board circuits that provide servo current feedback through an I2C bus back to the computing unit. Robot OSCAR-3
  • 30. 3.2 Hexapod Robots 15 doesn’t have a microcontroller onboard, but instead it is connected to a PC via a USB cable. It uses the “Generic robot architecture” [Gen08] concept in order to provide a better software driver access to the robot’s sensors and actuators, and therefore easier control and actuation of the robot. Fig. 3.8 Hexapod robot OSCAR-3. 3.2.2.4 Hexapod Robot Demonstrator – OSCAR - X The new prototype of the OSCAR robot generation, called OSCAR-X (Figure 3.9), is built to provide a better robot research test-bed for testing the biologically in- spired algorithms. In comparison to its predecessors, the OSCAR-X features a completely new design and was rebuilt from scratch. New features of the robot include: - Robot leg amputation mechanism: R-LEGAM [Jak09]; - Light weight glass-fiber body; - Robot legs spatially distributed in a circle with 60 degrees between each two neighboring legs; - Greater payload capabilities (sensors, batteries, camera, etc.) for the scientific measurements and experiments; - Stronger digital RX-64 servos with digital feedback for their real time posi- tions, torque levels, current levels, temperatures, etc. - Powerful Lithium-polymer batteries for the servos and electronics; - Weight of the body including the batteries is 7,5 kg; - Improved foot design for better detection of the ground, complete with binary contact sensors; - Powerful embedded system - Gumstix® “Verdex board” [GUM09] running embedded Linux; - Usage of the “Generic robotic architecture” [Gen08] concept, to provide bet- ter software driver access to the robot’s sensors and actuators and therefore easier control and actuation of the robot; - Orientation sensor; - Wireless camera and an additional camera servo.
  • 31. 16 3 Joint Leg Walking and Hybrid Robot Demonstrators (a) (b) (c) (d) Fig. 3.9 (a) Hexapod robot OSCAR-X in development stage; (b) OSCAR-X in nature; (c) Front view of robot OSCAR-X with onboard camera and additional ultrasonic sensors; (d) Top view of robot OSCAR-X. 3.2.2.4.1 Robot Leg Amputation Mechanism – R-LEGAM The main feature of the OSCAR-X is the improved design of robot’s legs, which aim for performing on-demand robot reconfiguration. Namely, the patented mechanism for robot leg amputation, R-LEGAM (DPMA-Az: 10 2009 006 934) [Jak09], is integrated for each of the OSCAR-X’s legs (Figure 3.9 and Figure 3.10). The robot’s leg can be detached from the robot’s body by software command. This is especially helpful when some of the legs malfunction. So instead of carry- ing the malfunctioned legs during the rest of the mission, the legs can be ampu- tated to prevent any other future negative influence on the rest of the functional robotic system. The in-situ reconfiguration of the hexapod robot OSCAR-X using biologically inspired approaches will be discussed in chapter 10.
  • 32. 3.3 Humanoid Robots 17 Fig. 3.10 (a) CAD design of Robot leg amputation mechanism: R-LEGAM; (b) R-LEGAM integrated on the robot’s body; (c) Robot’s leg detached from the robot’s body using the R- LEGAM mechanism. 3.3 Humanoid Robots Humanoid robots are robot demonstrators that have a human like appearance and are often used in robotics research, or as entertainment and service robots. Humanoid robots are therefore used to study and research the complexity of human walking and dynamic balancing, but also used for research in prosthesis development, human cognition, and human sensory information processing and perception. They have two legs, usually two arms and a head, and are equipped with lots of actuators and sensors including accelerometers, tilt sensors, cameras, pressure sensors on their feet, ultrasonic and infra-red sensors, etc. There are humanoid robot soccer matches organized by the RoboCup federa- tion [Rob10], where humanoid robots autonomously play soccer. Such competi- tions are important for the overall research in humanoid robots and emphasize the work on developing new algorithms for humanoid robot dynamic walking and stabilization, cooperation, localization on the field, etc. In chapter 6, research work is presented on self-stabilizing humanoid robot walking using biologically inspired algorithms. 3.4 State of the Art Humanoid Robots There are many humanoid robots known nowadays (March 2010) that are con- sidered state-of-the-art due to the number of features they exhibit. Here are some of the famous humanoid robots: Nasa’s “Robonaut 2” (R2) [NAS10], “TOPIO 3.0” [TOS09], “ASIMO” developed by HONDA [HON07], “Albert-Hubo” [HAN05] by Hanson Robotics, and “NAO” by Aldebaran Robotics [ALD10]. (Figure 3.11)
  • 33. 18 3 Joint Leg Walking and Hybrid Robot Demonstrators (a) (b) (c) (d) (e) Fig. 3.11 State of the art humanoid robots: (a) “Robonaut 2”; (b) “TOPIO 3.0”; (c) “ASIMO”; (d) “Albert-Hubo”; (e) “NAO”. The robot “Robonaut2,” nicknamed as “R2,” is a dexterous and technologically advanced humanoid robot developed by NASA and General Motors. The goal is to have this robot accompany the space missions and work side-by-side with humans. The “R2” has a torso equipped with a head and two arms but is without legs. “TOPIO 3.0” is the table tennis playing robot, designed and constantly im- proved by the company Tosio. The robot is said to use artificial intelligence algo- rithms to continuously improve its playing skill level. Robot “ASIMO,” developed by HONDA, is one of the most famous humanoid robots and is mostly used for entertainment purposes. This robot can detect faces, shake hands with humans, walk up and down the stairs, run, and even perform small jumps while running. Robot “Albert-Hubo” is built on “Hubo 2,” a “KHR-4” [HUB03] robot model, and is the next generation of the “KHR-3” humanoid robot with Albert Einstein’s head mounted on its body. The robot is used for artificial muscle actuator re- search, autism therapy, cognitive science research, etc.
  • 34. 3.5 Humanoid Robot Demonstrator - S2-HuRo (Self Stabilizing Humanoid Robot) 19 “NAO” is commercially available research humanoid robot platform equipped with a variety of actuators and sensors like: cameras, gyrometers, accelerometers, IR, and sonar sensors. It has a visual programming interface with which robotic movements can easily be developed. “NAO” robots are also in the RoboCup hu- manoid soccer games in the NAO - RoboCup standard league. Humanoid robots come in various sizes, ranging from small robots like NAO robot - 58 cm up to full size robots like TOPIO 3.0 - 188cm size (Figure 3.11). 3.5 Humanoid Robot Demonstrator - S2-HuRo (Self Stabilizing Humanoid Robot) The S2-HuRo robot was developed for the research on biologically inspired tech- niques for humanoid robot stabilized walking, presented in chapter 6. S2-HuRo is based on the “ROBONOVA-1” (Figure 3.12) [HIT06] robot kit with “HSR- 8498HB” digital servos, “MRC-3024” servo control, and integrated I/O board. Fig. 3.12 Humanoid robot “ROBONOVA-1” This robot has been modified extensively and specially tuned, taking the final form and construction as shown in Figure 3.12. The modified robot is called the S2-HuRo (Self Stabilizing Humanoid Robot) presented in Figure 3.13 (a)-(d). The robot is built of the following hardware ele- ments: embedded system - Gumstix® Verdex [GUM09] represented in Figure 3.14 with wireless module and antenna, 2-axis accelerometer and gyroscope, Lith- ium-polymer battery pack, voltage converters for 3, 5, and 6 volts for the electron- ics, and binary sensor contacts (Figure 3.15). In order to decrease the weight of the robot, some servos from the “Robonova 1” arms have been removed during the robot hardware tuning, and the connection between the segments at those points is made with screws instead. The final S2-HuRo robot appears as seen in Figure 3.13.
  • 35. 20 3 Joint Leg Walking and Hybrid Robot Demonstrators (a) (b) (c) (d) (e) Fig. 3.13 (a) – (d) S2-HuRo (Self Stabilizing Humanoid Robot). Fig. 3.14 Gumstix® Verdex embedded system with wireless LAN module, antenna, MMC card, and additional serial connector cable. View with three sensors per foot. “L” and “R” indicate the left and the right robot’s legs.
  • 36. 3.5 Humanoid Robot Demonstrator - S2-HuRo (Self Stabilizing Humanoid Robot) 21 When comparing the robots shown in Figure 3.12 and Figure 3.13, differences in robot modifications can easily be spotted. By looking at the top view of the ro- bots, the robot (Figure 3.13) (c) the two axis accelerometer sensors can be spotted under the robot’s head. In the backside figures (Figure 3.13) (d) and (e) of the S2- HuRo, the case of the embedded Gumstix® Verdex system can be seen along with the wireless antenna next to the case. On the robot’s feet Lithium-polymer batteries can be spotted – two batteries per foot. The relatively light weight 5Wh LiION rechargeable batteries used by S2-HuRo for servo and electronics power supply are the same batteries used by E-pucks robots [EPU09]. The voltage from the batteries is further down-regulated by integrated voltage convertors to be compatible with the voltage operating range of the servos, Microcontroller AtMega “MRC-3024” board, and the additional electronics. The batteries are located on the robot’s feet so the center of gravity of the robot is lowered and this increases the dynamic stability of the robot. Fig. 3.15 Schematic view of binary contact sensors by the S2-HuRo feet – bottom.
  • 38. B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 23–34. springerlink.com © Springer-Verlag Berlin Heidelberg 2011 Chapter 4 Biologically Inspired Robot Control Architecture Robot control architectures are related to sensing, monitoring, and acting actions of the robots. They are an important part of each robot control and the coordina- tion of their behaviors. There are different kinds of robotic architectures imple- mented by different kinds of robots. Most of the control robot architectures are divided into following groups: - Reactive and Subsumption / Behavior-based control architectures (schema based); - Deliberative control architectures (hierarchical) or Sense-plan-act architec- tures; - Hybrid control architectures. These “common” types of robotic architectures are explained in the subsection that follows. On the other hand, the research of biologically inspired robot control architec- tures aims at developing robot control architectures that would have life-like properties and are able to self-organize their constitutive components instead of predefining them manually. They must be able to self-optimize for their best performance and be capable of detecting the newly attached components such as sensors and actuators. They must also be able to autonomously re-configure them- selves and continue with execution of their mission tasks. Although these are all nice features to have for robot control architecture, de- veloping such “life-like” robot control architecture is rather complicated. The tricky design of such architecture involves re-thinking aspects such as: - The types of elements that would constitute such control architecture; - The property of the architecture must be generic and independent of the configuration and attached hardware and electronics; - How can an on-demand reconfiguration be achieved in such autonomic ar- chitecture so that the new components can be easily added to the running system? - Where to place the “main control” unit or if such a unit should exist at all in such module-based architecture.
  • 39. 24 4 Biologically Inspired Robot Control Architecture Questions also arise on how the decision for a particular detailed design of such architecture might influence the proper working of the other constitutive elements of the architecture. Properties such as scalability and emergence also have to be addressed. 4.1 Overview on “Standard” Types of Robot Control Architectures In this section several common types of robot control architectures are introduced: Reactive control architecture; Subsumption control architecture; and Delibera- tive/reactive control architecture. Then an overview on autonomic types of architectures is given, before an explanation of the ORCA (Organic Robot Control Architecture). ORCA is on the other side directly related to the research presented in this book as well as to several ideas that are presented about improvement of robot control architecture using biologically inspired concepts. 4.1.1 Reactive and Subsumption and Behavior Based Control Architecture By reactive control architectures (schema based), the control is stimulus-response based, where behaviors are represented by direct sensor to actuator predefined reaction mappings. (Figure 4.1) Although the speed of response by the reactive control architecture is rather high, which might be suitable for some real world scenarios where the reaction time might be very important, reactive architectures are perhaps not suitable for tasks where predictive planned outcomes should be generated. Fig. 4.1 Principle of reactive control architecture. An alternative approach to the reactive system architecture is subsumption architecture (behavior based) introduced by Brooks in 1986 [BRO86]. This ap- proach is based on priority behaviors organized into layers, where higher priority behaviors subsume lower priority behaviors. In subsumption architecture, the low- er layer behaviors (reflexes) can inhibit higher layer behaviors. In this bottom-up fashion, the reflexes are bottom layers that can be expressed in the fastest way. The upper layers are related to higher robot control work in inhibited fashion, de- pending on their priority and currently executed lower actions. It is also important to mention that there is no higher level supervision in this architecture. The subsumption architecture can be useful when overall robot behavior should be
  • 40. 4.1 Overview on “Standard” Types of Robot Control Architectures 25 dynamic, reactive, emergent. Difficulties in this architecture are related to deter- mining the priorities of the layers constituting the architecture. An example struc- ture of subsumption architecture is presented in Figure 4.2. Fig. 4.2 An example structure of subsumption architecture. 4.1.2 Deliberative Control Architecture Deliberative control architectures are based on the Sense-Plan-Act principle. For their optimal functioning, deliberative hierarchical control architectures usually need full knowledge about the environment. By these types of control architectures, the robot first senses the environment, then plans potential solutions and considers the results when choosing appropriate actions. It is assumed that the world model is provided. The robot then executes the actions through actuators. The structure of such architecture is represented in Figure 4.3. Fig. 4.3 Sense-Plan-Act model of deliberative control architecture.
  • 41. 26 4 Biologically Inspired Robot Control Architecture The advantages of using the deliberative architecture is that in such goal ori- ented control architecture the goal of a given task can be achieved in a planned way. However, the difficulties or drawbacks of using such architecture are related to the re-planning phases which introduce slow response to some actions. There- fore the architecture is perhaps not suitable for tasks where fast response is needed. An additional drawback is that in case the environment changes, there also need to be changes in the control architecture, so its reaction can be compati- ble with the changed model of the environment. 4.1.3 Hybrid Control Architecture One way of mitigating the limitations and drawbacks seen by the reactive and the deliberative control architectures is to combine both of the architectures into a hybrid control architecture. A first proposal for usage of such hybrid architecture was made by Arkin [Ark87]. Since then, different kinds of hybrid robot architectures have been pro- posed [Con92] [LHG06]. In general, the hybrid architecture uses higher level planning in order to guide the lower level of reactive components. It is often depicted as a three layer architecture, where the top layer is the deliberative layer, operating under a slower sampling rate than the bottom layer, which is the reactive layer with a fast reaction time. The middle layer might have different interpretations and implementations for different projects, for example, aggrega- tion of information coming from the lower layer. This control architecture can be represented as in Figure 4.4. Deliberative layer Intermediate layer Reactive layer Fig. 4.4 Model of a hybrid control architecture. The benefit of using such architecture is that it is still a goal oriented architec- ture where planning for the next actions occur by a deliberative layer, and at the same time “lower level” actions can be executed by the reactive layer. Therefore it can be more or less assured that the proper planned actions will be executed
  • 42. 4.2 Overview on Autonomic Control Architecture 27 and also that the robot will interact better and react faster to changes in the environment. 4.2 Overview on Autonomic Control Architecture Previously introduced robot control architectures (reactive, deliberative, and hybrid architectures) have the modules, behaviors, and tasks mostly planned, modeled, and defined in advance by a human operator. Autonomic control archi- tectures use the idea to build architectures that will easily cope with the high complexity of the technical systems, and that dynamically adapt with respect to available resources and user needs [NaB07]. Responses taken automatically by a system without real-time human intervention are called autonomic responses [SHR06] [LVO01]. Given only high-level commands, the autonomic systems should be able to manage themselves [KeC03] in a self-governing manner. The idea of autonomic computing was first introduced by IBM in their Manifesto for Autonomic Com- puting [IBM01]. They proposed several features that autonomic systems should exhibit such as: self-configuration, self-healing, self-optimization, and self- protection, all of which were inspired by the human body’s autonomic nervous system. These terms were explained previously in Chapter 2.4. Autonomic systems consist of autonomic elements – which can build relation- ships with other autonomic elements and manage and influence or change their behavior in order to comply with the higher level policies defined by human op- erators. Such an autonomic element is represented in Figure 4.5. As seen from the figure, each autonomic element has an autonomic manager and one or more managed elements. An Autonomic manager is associated with control of the autonomic element and actions like Monitoring, Analysis, Planning, Execution using a Knowledge base. Managed elements on the other hand repre- sent the hardware resources like storage, the CPU, etc. The autonomic manager controls or influences the execution of the Managed element and monitors its operations. Therefore it is a closed feedback loop architecture. The concept and functionality of the autonomic control architecture differs from the “common” control architectures, and introduces the notions of self- configuration, self-healing, self-optimization, and self-protection, working towards building self-managing systems.
  • 43. 28 4 Biologically Inspired Robot Control Architecture Fig. 4.5 Structure of autonomic element [KeC03]. Fig. 4.6 Generic Observer/Controller architecture.
  • 44. 4.3 ORCA (Organic Robot Control Architecture) 29 Within the Organic Computing initiative an Generic Observer/Controller archi- tecture (Figure 4.6) has been firstly introduced in [RMB06], which incorporates Observer and Controller units, where the Observer monitors the proper behavior of Controller units and modifies their control in order to suit the predefined goals. Observer/Controller architecture is a closed feedback loop architecture. 4.3 ORCA (Organic Robot Control Architecture) ORCA development [BMM05] is a result of Organic Computing (Chapter 2.2) research on developing hybrid robust robot control architecture that has self-x properties and at the same time provides safe and reliable functioning. Usually it is important for control architectures that exhibit self-x properties to have the controlled emergence property. Namely, the system should be able to learn and adapt its behavior, but at the same time not demonstrate some unwanted behavior that exceeds some pre-defined constraints defined in the system’s core specifications. Fig. 4.7 ORCA – Organic Robot Control Architecture. ORCA architecture is therefore built to satisfy these criteria, to provide reliable robot function control, to have modular architecture design, and to allow for emergent properties of its constituent BCU and OCU units.
  • 45. 30 4 Biologically Inspired Robot Control Architecture Basic Computing Units (BCUs) are basic software modules in the ORCA archi- tecture, which may implement different functionalities related to robot control or the robot’s hardware. These functionalities can be related to: sensor values acqui- sition, sensor data fusion, sensor information pre-processing, or for example to control a robot leg segment. Organic Computing Units (OCUs) are special type of units in the ORCA architecture related to monitoring tasks for the correct behav- iour of BCUs and also to control them to provide counteractions in case of anoma- lies. ORCA architecture is presented in Figure 4.7. ORCA is closely related to the research presented in this book, as a concept of a robot control architecture. 4.4 Distributed ORCA Architecture for Hexapod Robot Control ORCA has also been adapted to suit the practical experiments conducted on experimental robots OSCAR-2 and OSCAR-X. control signals feedback signals Fig. 4.8 Decentralized ORCA used in several robot experiments.
  • 46. 4.5 Cell Differentiation as Biological Inspiration for Enhanced ORCA 31 The idea behind practically implemented distributed organization of ORCA is that each leg represented in ORCA consists of a number of BCU units which can control some servo movements, and are related to some behaviors such as leg gait generation, swing/stance leg movements, etc. There can be one OCU unit that is responsible for monitoring the health status of the BCU units and providing coun- teractions in case of detected anomalies. There is one OCU unit per leg, and there- fore several of them in the whole architecture related to monitoring the behaviors and health status of the leg units. On the other hand, for anomaly detection purposes, a distributed ORCA can be considered with one OCU per each BCU related to servo movement. Depending on the application and task that the robot must realize, different types of distrib- uted control architectures can be considered. For example, the distributed ORCA architecture represented in Figure 4.8 has been used in the experiments with emergent robot walking with distributed pres- sure on its feet and for firefly inspired self-synchronization of walking gait of the hexapod robot. In the following chapter information is given on exploring new ideas about ORCA architecture that would improve the robust and autonomic functioning of this control architecture. 4.5 Cell Differentiation as Biological Inspiration for Enhanced ORCA An initial biologically inspired research was done on further enhancing the stan- dard ORCA architecture, in order to provide means for achieving the self-x prop- erties, like the self-organization of the modules by the robot control architecture. A biological phenomenon called cell differentiation has been explored, more specifically the self-organization that happens in this process. This study may aid in improving the functionality of the ORCA control architecture. First, a quick overview of the biological concept cell differentiation is appro- priate before exploring the ideas of how these biological phenomena might be beneficial for ORCA. 4.5.1 Overview of a Biological Concept – Cell Differentiation By biological systems, the self-organization of the components seems to be intrin- sically integrated. This can be seen when observing the cells in the organs and tissues. Namely, the cells in one organism generally have some common charac- teristics that can be observed in all the cells in the organism. For example, they all carry the same DNA. But on the other hand, cells also do differ from each other. If we observe the brain cells, muscles cells, liver cells, etc., they all differ in their function or inner structure. Research on types of cells called stem cells [BMT63] [SMT63] has given in- sight into types of cells that are able to develop into different types of cells in the
  • 47. 32 4 Biologically Inspired Robot Control Architecture body. Those stem cells are totipotent or pluripotent, and when dividing can poten- tially (depending on the environment) transform and become any type of cell like brain cell, muscle cell, liver cell, etc. This is done in the process of cell differentia- tion [STE10], in which the cell due to various environmental conditions, inter- signaling between the cells, or physical contact with neighboring cells can start to develop into a specific type of cell. 4.5.2 The Enhanced “Stem” Type ORCA Architecture If we have a look into ORCA and its OCU and BCU units and their pre-defined functionality, we might consider these constitutive units as some sort of “already differentiated cells” in the architecture. By ORCA structure mentioned earlier, some BCU units can be related to motor control; some to control segments of the robot; others can be related to provide sensor values read from the sensors, etc. The OCUs can on the other hand be related to monitoring the individual BCU units. The idea behind utilizing the cell differentiation concept in this context would be on introducing some “stem cell” like types of units in the architecture that can “differentiate” into appropriate module types. This would be especially useful for situations where other components like some servos and actuators (connected via bus), should be dynamically installed into the robotic system without any need of human operator intervention. In an ordinary case such operation would require that the human operator iden- tifies the type of module that has to be incorporated into the architecture and then programs its interface so that the component can be suitably accessed. For exam- ple, if it is an actuator, then it should be newly interfaced into the robotic system and should receive commands for actions via its interface. If it is a sensor, then it should provide data through its interface to the units that need it. On the other hand, by the robot control architecture that has “stem” type of units, it will not have them all preprogrammed and a fixed topology of the units’ interconnection. Some units at the start might be defined, for example some BCU units controlling motors in the robot. But if any additional motors are added to the robotic system, then by “fingerprint” of their functioning, they might appear to have similarities with the already defined BCU units for the motor control. In that case those “stem” type of units can be associated with BCU type interface for mo- tor control. Analogously this can be done also for any other extra sensors connected to the robotic system. If a newly connected sensor device has a “fingerprint” that might be similar in functionality to that of BCU units existing in the robotic system and related to sensor signal acquisition, then such extra added sensor units will get associated with BCU sensor type interface. At first, all newly inserted hardware units are associated with some “stem” type of units in the control architecture, and then they “differentiate” within the archi- tecture to unit types in the control architecture that best describes their functional- ity. When speaking of OCU units in the context of such enhanced “stem” ORCA architecture, the “stem” type of OCU units might differentiate into other different
  • 48. 4.5 Cell Differentiation as Biological Inspiration for Enhanced ORCA 33 types of OCU units associated with monitoring of proper operation of associated BCU units. Such “stem” type of BCU and OCU units might be interesting to explore, since upon replication and differentiation the newly generated and associated units will perhaps express new intelligent functionality of the robotic system. On the other hand these experiments with “stem” type of BCU and OCU units should be ap- proached with great care, since there is also an open possibility that by such dif- ferentiation the OCU units might expose an undesirable property or behavior of the robotic system. There may be other techniques introduced that help to prevent this from happening, like introducing special OCU monitor units that will “guide” this process of differentiation to be compatible with the pre-defined functional requirements of the robotic system. The enhanced “stem” ORCA architecture model is presented in Figure 4.9. Fig. 4.9 Enhanced “stem” type ORCA. The whole mechanism of differentiation of the BCU and OCU units should be explored, in the sense of defining a method for converting the “stem” type BCU or OCU units. For example, they may be influenced by the neighboring units to be- come a completely defined unit within the control architecture with a specific function and interface.
  • 49. 34 4 Biologically Inspired Robot Control Architecture One idea would be that the data oriented sequences originating from the OCU and BCU units can be compared to the data streams originating from new “stem” units. This might help the “stem” OCU or BCU units to be influenced to change their type to that kind of BCU or OCU units that have similar data streams. This would be analogue to the biological counterpart of cell differentiation, where the different types of cells have or release slightly different types of molecules near their vicinity, which differ from the molecules released in their surroundings by other type of cells in the organism. By the enhanced “stem” type ORCA organized in a bottom-up fashion, the dif- ferentiation that occurs by the “stem” type of lower level BCU units and lower level OCU monitoring units, is related more to the reactive behavior of the system. In contrast, the higher level differentiated BCU and the higher level OCU units (provided for their monitoring) are related to expressing the cognitive behavior of the system. The rectangles show an example on how such elements would com- municate with each other in such enhanced “stem” type ORCA.
  • 50. B. Jakimovski: Biologically Inspired Approaches, COSMOS 14, pp. 35–66. springerlink.com © Springer-Verlag Berlin Heidelberg 2011 Chapter 5 Biologically Inspired Approaches for Locomotion of a Hexapod Robot OSCAR Different types of walking gait generation approaches have been considered for walking by various multi-legged robots. Some of them are based on mathematical formulations [PaH09] or inverse kinematic models [ShT07] trying to mathematically model and describe the kinematics of the robot movements and also the interaction of the robot with the environment. This may prove difficult since completely modeling the robot and its interaction with the environment and other environmental influences on proper working of robotic components is very complex. Another kind of approaches are biologically inspired. The biologically inspired approach CPG (Central Pattern Generator) is based on walking patterns generated by neural-networks [Mat87] seen by animals and insects. This is acquired for producing walking movement by multi-legged walking robots [ITG09]. Another type of biologically inspired approaches can exhibit self-x properties such as: self-organization or self-reconfiguration, similar to self-organization properties seen in biological systems. Before describing in detail the research done about such self-organizing walking gait patterns based on emergence, several characteristics are given for locomotion by insects and CPG based types of “common” walking gaits seen by animals and insects. 5.1 Characteristics of Locomotion Seen by Insects and Animals - Applied to Robotics Domain 5.1 Characteristics of Locomotion Seen by Insects and A nimals Observations and research done on insects, arthropods and animals has provided new insights on the locomotion seen in nature and on how living organisms generate their movement patterns. These observations have been useful for developing basic insect movements. For example, research on the stick insect (Carausius morosus) [Cru76] walking has given new information about the functionality of the leg segments and their relation to protraction and retraction; elevation and depression; extension and flexion.
  • 51. 36 5 Biologically Inspired Approaches for Locomotion of a Hexapod Robot OSCAR There are two phases that are characteristic and essential for leg movement by the insect: swing and stance phases. There are also two remarkable positions for these movements – the posterior extreme position (PEP) – the position of the leg on the ground when the leg is at the end of the power stroke; and the anterior extreme position (AEP) – the position of the leg on the ground when the leg is at the end of its return stroke. In the swing phase, the leg is moving from the PEP to AEP, shown in Figure 5.1.1 – movement which precedes the movement of the leg over the ground. In the stance phase, the leg is moving from AEP to PEP, which produces the thrust that moves the insect over the ground. The swing and stance phases are characterized with the length of their respective trajectories. The lengths of swing and stance trajectories have direct influence on the speed with which the insect is moving over the terrain. The longer the trajectories of swing and stance, the bigger the distance travelled by the leg on the ground, and vice-versa, the shorter the trajectories, the shorter the distance travelled on the ground. The length of swing and stance phases by one leg in combination with swing and stance phases of other legs may influence the insect to turn around its vertical axis. As represented with circles in Figure 5.1, the leg of the insect can be considered to be a 3 degrees of freedom (3 DOF) system. Namely, the leg consists of three segments connected via joints, providing basic movements for protraction and retraction; elevation and depression; extension and flexion. Fig. 5.1 Swing and stance phases of an insect’s leg. The same principle seen by leg structure and movement by insects has been used for constructing and providing locomotion of the legs of multi-legged walking robots. In Figure 5.2 (a), a 3 DOF structure is indicated by one of the robot’s legs, which is similar to structure seen by the insect’s leg – the circles represent the joints and the degrees of freedom. Servo “Alpha” is the nearest to the body on each of the legs; servo “Beta” is next to it, down the leg; servo “Gamma” is the last servo on the leg outward from the body. The swing and stance phases and their trajectories by robot’s leg are presented in Figure 5.2 (b), which are similar to swing and stance phases seen by an insect’s leg (Figure 5.1).
  • 52. 5.2 Central Pattern Generators (CPG) 37 (a) (b) Fig. 5.2 (a) 3 DOF structure represented with circles on one of the robot’s legs; (b) Swing and stance phases and their trajectories of the robot’s leg. 5.2 Central Pattern Generators (CPG) Research on walking gait patterns has shown that neural networks called Central Pattern Generators or CPG are located in the neural systems below the brain stem [Gri81] in the spinal cord and are responsible for generating and modulating the walking patterns and others specific to rhythmic motions. Mathematical models for CPG are proposed in [Mat87] where the CPGs consisted of two neurons, where each neuron receives an excitatory tonic input, a mutual inhibition, and an external inhibitory input. At the end the output is SELF INHIBITION FROM OTHER OSCILLATOR TONIC EXCITATION FEEDBACK OUTPUT MUTUAL INHIBITION + - Fig. 5.3 CPG model with two inhibiting neurons.
  • 53. 38 5 Biologically Inspired Approaches for Locomotion of a Hexapod Robot OSCAR generated as a joint interaction of such inhibiting neurons. The mutual self- inhibition is useful to induce a stable relaxation without any external input, therefore an oscillation can be provided. Figure 5.3 represents the schematic of such a model of two inhibiting neurons. The neurons which mutually inhibit themselves can be related to flexion or extension movements of the legs, which are important for walking. It has been also shown that activities of CPGs can be modified by sensory feedback [CoB99] and reflexes, which gives insight on how the rhythmic and reflex movements can be coupled. Also Holk Cruse together with Friderich Pfeiffer did pioneer work on a better understanding of biological control systems for 6 legged walking machines (as well as Büschges, Berns, Ilg, Albiez,..). 5.2.1 Common Observed Gaits by Insects Several walking gaits, which might be generated by an insect’s CPGs have been observed: Wave gait, Ripple gait, and Tripod gait. - Wave gait – is the slowest gait where one leg is in swing phase (in the air) while the other legs are in the stance phase (on the ground). This gait is characterized as the most stable one, since all the other legs are on the ground and supporting the robot’s body. However, this is also the slowest walking gait, since only one leg is in the air at a time, while the others are on the ground. - In Ripple gait – there are two independent gaits from both sides of the body. The stance phase is usually double the swing phase and the opposite legs are 180 degrees out of phase. - The research on walking gaits of six legged insects, for example a cockroach [SpM79], has indicated that the Tripod walking gait provides the six-legged insect with the fastest speed over the ground. Tripod walking gait is a gait by which at any moment of time three of the robot’s legs are in the swing phase, while the other three legs are in the stance phase. The three legs on the ground provide the insect with static and dynamic stability while walking. These gaits are represented in Figure 5.4 (a), (b), (c), going from the slowest one – wave gait, then ripple gait, and lastly the fastest: tripod gait. The black filled bars indicate the stance phase, while the non-filled ones represent the legs during swing phases. On the symbolic represented insect in Figure 5.4, the symbols L1, L2, L3 indicate the legs on the left side of the body with their respective numbering and position on the body. The symbols R1, R2, R3 indicate the legs on the right side of the body with their respective numbering and position on the body. The symbols for left and right legs and their numberings are located on the vertical axis, so it can be seen for which legs what type of leg movement is taking place. The horizontal axis represents the time domain.
  • 54. 5.3 Experiments with Self-organizing Emergent Robot Walking Gait 39 Wave gait (a) Ripple gait (b) Tripod gait (c) Swing phase Stance phase R1 R2 R3 L1 L2 L3 Time Fig. 5.4 Common gaits observed in insects (adapted from [FaC93]). The presented insect walking gaits have been successfully applied for generating robot walking for the domain of multi-leg walking robots patterns [IYS06] [YAL06] [MDB07]. The fastest walking gait, tripod gait, has often been applied to robotic projects to generate walking motion by the six legged walking robots. The tripod gait has also been used within this research to provide the walking for the robot OSCAR-X in some of the experiments which will be explained later in chapter 5.4. 5.3 Experiments with Self-organizing Emergent Robot Walking Gait with Distributed Pressure on Robot’s Feet 5.3 Experiments wit h Self-organizing Emergent Robot Walking Gait In the previous chapters 2.4 and 2.5, the terms about self-x properties and the emergence were explained as being important for exploration and implementation
  • 55. Other documents randomly have different content
  • 56. with minute decumbent hairs, less easily seen but still existing in the secondary pair. In the Hymenoptera in general the wings are covered with minute hairs or bristles; but in Tiphia, Scolia—with the exception of S. Radula and affinities in which they are hairy—and others, the wings are nearly naked; in Pompilus, Pepsis, c., the hairs are infinitely numerous and very short; in the Sphecidæ, Mutilla, c., they are more distinct, longer, and less numerous; in the humble-bee (Bombus) and many others the apex of the wing is darkened by a large number of more conspicuous hairs, each of which seems to spring from a minute tubercle: as these tubercles are in a part of the wing that is strengthened by few nervures, they may probably be intended to supply their place, in giving firmness and tension to this part. The wings of Diptera, under the present head, may be viewed with regard to the hairs that are implanted in the membrane of the wing, in its nervures, and in its margin. In the first view, in Stratyomis and immediate affinities the wing is nearly naked; but in Xylophagus, Beris, and the great majority of the Order, the membrane of the wings is thickly planted with innumerable very minute bristles, not to be seen but under a powerful lens, often black, and seemingly crowning a little prominence, and giving the wing an appearance of the finest net-work. As to the clothing of the nervures, the costal, in Anthrax, Bombylius, c., is often remarkably bristly at the base, with hairs intermixed; in Œstrus Ovis, in the inner margin or edge of this nervure, is a single series of bristles, or rather short spines, like so many black points; in Œ. Equi the whole costa is covered with short decumbent hairs or bristles; in Musca pagana F., just at the apex of the costal areolet, that nervure is armed with a spur or diverging bristle larger than the rest, which is also to be found in many others of the Muscidæ, some of which have two and others more of these spurs. The little moth-like midges (Psychoda Latr., Hirtæa F.) at first appear to have the whole surface of their wings covered with hairs; but upon a closer examination it will be seen that they are planted in the nervures, from each of which they diverge, so as under a lens to give it a very elegant appearance[1956]. This fly has its wings beautifully fringed with fine
  • 57. hairs, the third circumstance to be attended to under this head; in the Tipulidans, and many others of this Order, the apex and posterior margin are also finely fringed with short hairs. Some Dipterous insects make a near approach to the Lepidoptera in the covering of their wings: in the common gnat, when the wings are not rubbed, the nervures are adorned by a double series of scales, and the marginal fringe also consists of them[1957]; and in a Georgian genus, which appears in some degree to connect Culex with Anthrax c., there are scales scattered upon the membrane as well as upon the nervures; besides, its antennæ[1958] and abdomen are also covered with them. The Order, the clothing of whose organs of flight excites the admiration of the most incurious beholder, is that to which the excursive butterfly belongs, the Lepidoptera. The gorgeous wings of these universal favourites, as well as those of the hawk-moths and moths, owe all their beauty, not to the substance of which they are composed, but to an infinite number of little plumes or scales so thickly planted in their upper and under surface, as in the great majority entirely to conceal that substance. Whether these are really most analogous to plumes or scales has been thought doubtful. De Geer is inclined to think, from their terminating at their lower end in little quills and other circumstances, that they resemble feathers as much as scales[1959]; Reaumur on the contrary suspects that they come nearer to scales[1960]. Their substance, approaching to membrane, seems to make further for the former opinion, and their shape and the indentations that often occur in their extremity, furnish an additional argument for the latter. Their numbers are infinite; Leeuwenhoek found more than 400,000 on the wings of the silk-worm moth (Bombyx Mori)[1961]; and in those of some of the larger moths and butterflies the number must greatly exceed this. You will observe however that in many Lepidoptera the wings are partially, and in some instances generally, transparent: thus in Hesperia Proteus, a butterfly before noticed for the long tail that distinguishes its secondary wings, there are many transparent spots;
  • 58. in Attacus Atlas, one of the largest of moths, and its affinities, there is as it were a window in each wing formed by a transparent triangular space; in A. Polyphemus, Paphia, c., the pupil of the ocellus is transparent, which in the former is divided by a nervure. In several of the Heliconian butterflies, and in Zygæna F., c., the greater part of both wings is transparent, with scales only upon their nervures, round their margin, or forming certain bands or spots upon them; in Parnassius Apollo, Mnemosyne, c., the scales are so arranged as not wholly to cover the wings, which renders them semidiaphanous; and in some (Nudaria) the wings are intirely denuded. With regard to size, the scales vary often considerably in different tribes; in Heliconia they appear to be more minute than in the rest; and in Castnia they are the largest and coarsest; the extremity of the wings of Lepidopterous insects in general is fringed with longer scales than their surfaces, and even those of the last in the same wing; sometimes vary in magnitude. The little seeming tooth that projects from the middle of the posterior margin in the upper wings of Notodonta, a subgenus of Bombyx L., is merely produced by some longer diverging hairs. The shape and figure also of scales are very various—some being long and slender; others short and broad; some nearly round; others oval, ovate, or oblong; others spathulate; others panduriform or parabolical; some again almost square or rhomboidal; many triangular; some representing an isosceles triangle, and others an equilateral one; lastly, some are lanceolate and others linear; again, some have a very short pedicle and others a very long one: with regard to their extremity; some are intire, without projecting points or incisions, while others are furnished with them: of these some terminate in a single long mucro, others have several shorter ones; some are armed with teeth, varying in number from two to thirteen in different species[1962]. Many other forms might be enumerated, but these are sufficient to give you a general notion of the infinite variety of this part of the works of the Creator. I must next say a word or two upon their arrangement on the wing. In most instances this is in transverse lines, which sometimes vary a little from a rectilinear
  • 59. course, and the extremity of the scales of one row reposes on the base of those of the succeeding one, so that in this respect their arrangement is like that of tiles in a roof: in some cases it is not so regular: thus the minute scales on the wings of Parnassius Apollo, and others with subdiaphanous wings, are arranged without order; in Pieris and other Diurnal Lepidoptera, and many of the Crepuscular and Nocturnal, there appears to be a double layer of scales on both sides of the wing; the under layer usually consisting of white ones. If you denude the wings of any butterfly, which you may easily do by scraping it lightly on both sides with a penknife, you will be amused to trace the lines in which the scales were planted, consisting of innumerable minute dots: the lines of the under side, in some cases, so cut those of the upper side, as by their intersection to form lozenges. With regard to the position of the scales on the wing, they usually lie flat, but sometimes their extremity is incurved: in the beautiful Argynnis Vanillæ a very singular appearance of numerous transverse ridges is produced by the extremity of those scales that cover the longitudinal nervures of the primary wings, except at the base, being recurved. But though the general clothing of the wings of Lepidoptera consists of these little scales, yet in some cases they are either replaced by hairs or mixed with them. Thus, in the clear parts of the wings of Heliconians, Attaci, c., short inconspicuous hairs are planted; in a large number of the Orders the upper side of the Anal Area of the secondary wings is hairy; in several Crepusculars (Sphinx Phœnix, c.), where there is a double layer as before mentioned, the upper one consists of dense hairs, except at the apex, and the lower one of scales; and in most of them the scales of the primary wings are piliform, and the secondary are covered by what approach very near to real hairs; many of the Attaci are similarly circumstanced: the four wings of A. Cytherea are also covered externally with hair. 7. Before I conclude this long diatribe on the organs of flight of insects, I must not omit some notice of the infinite diversity of colours with which their wings are often variegated and adorned by
  • 60. the Creator, who loves to delight us by the beauty, as well as to astonish and awe us by the immensity and grandeur of his works. Though the wings in every Order exhibit instances of brilliant and beautiful colouring, yet those of the Lepidoptera in this respect infinitely excel them all, and to these, under this head, after noticing a few in the less privileged Orders, my observations will be confined. Although in the Coleoptera the wings are seldom distinguished by their splendour; yet those of some Cetoniadæ, as Cetonia africana, are extremely brilliant, and resemble those of many Xylocopæ in the lovely violet hue that adorns them: amongst the Orthoptera some Pterophyllæ, and in the Homopterous Hemiptera some Fulgoræ, emulate the Lepidoptera in the ocelli that give a kind of life to these organs[1963]; and a vast number of the destructive tribe of locusts (Locusta Leach) are remarkable for the fine colours and gaiety of their wings[1964]; in the Neuroptera numerous Libellulinæ emulate the Heliconian butterflies by their maculation; and in the genus Ascalaphus, which represents the Lepidoptera by its clubbed antennæ[1965], many also have the resemblance increased by the painting of their wings, so that some Entomologists have actually considered some of them as belonging to that Order[1966]; the wings of the Xylocopæ, before alluded to, sometimes add to the deep tints of the violet—which also prevail in the wings of several Diptera— towards their extremity the most brilliant metallic green or copper varying, As the site varies in the gazer's hand, and even those wings that consist of clear colourless membrane are often rendered extremely beautiful from the reflection of the prismatic colours. I should undertake an endless task did I attempt to specify all the modes of marking, clouding, and spotting, that variegate a wing, and all the shades of colour that paint it, amongst the Lepidopterous tribes; I shall therefore confine myself to a few of the principal, especially those that distinguish particular tribes and families. Of whole coloured wings—I know none that dazzle the eye
  • 61. of the beholder so much as the upper surface of those of Morpho Menelaus and Telemachus: Linné justly observes that there is scarcely any thing in nature that for brightness and splendour can be paralleled with this colour; it is a kind of rich ultramarine that vies with the deepest and purest azure of the sky; and what must cause a striking contrast in flight, the prone surface of the wings is as dull and dark as the supine is brilliant, so that one can conceive this animal to appear like a planet in full radiance, and under eclipse, as its wings open and shut in the blaze of a tropical sun: another butterfly, Papilio Ulysses, by its radiating cerulean disk, surrounded on every side by a margin intensely black, gives the idea of light first emerging from primeval obscurity; it was probably this idea of light shining in darkness that induced Linné to give it the name of the wisest of the Greeks in a dark and barbarous age. I know no insect upon which the sight rests with such untired pleasure, as upon the lovely butterfly that bears the name of the unhappy Trojan king (P. Priamus); the contrast of the rich green and black of the velvet of its wings with each other, and with the orange of its abdomen, is beyond expression regal and magnificent. But peculiar beauties of colour sometimes distinguish whole tribes as well as individuals. What can be more lovely than that tribe of little butterflies that flit around us every where in our summer rambles, which are called blues, and which exhibit the various tints of the sky? Lycæna Adonis of this tribe scarcely yields to any exotic butterfly in the celestial purity of its azure wings: our native coppers also, Lycæna dispar[1967], Virgaureæ, c., are remarkable for the fulgid colour of these organs; in Argynnis the upper side of their wings is tawny, spotted with black, while the under side of the secondary ones is very often adorned by the appearance of silver spots. How this remarkable effect of metallic lustre, so often reflected by spots in the wings of butterflies, is produced, seems not to have occupied the attention of Entomologists. M. Audebert is of opinion that the similar lustre of the plumes of the humming birds (Trochilus) is owing to their density, to the polish of their surface, and to the great number of little minute concave mirrors which are observable on their little
  • 62. beards[1968]. But these observations will not apply to the scales of the wings of butterflies, which are always very thin and generally very flat: in some instances, as in Morpho Menelaus, there appears more than one very slight channel upon a scale; but this takes place also in others that reflect no lustre. Their metallic hues must therefore principally be occasioned by the high polish of their surface and the richness of their tints. It is the purity of the white, in conjunction with their shining surface, contrasted with the dull opaque colour of the under side of the secondary wings, that causes the spots that decorate those of the Fritillaries (Argynnis) to emulate the lustre of silver. In Papilio the Trojans are distinguished by the black wings with sanguine spots, and the Greeks by the same with yellow spots; but these have proved in some instances only sexual distinctions[1969]. In the Danai candidi L. the colour of the tribe may be described as sacred to the day, since every shade, from white or the palest yellow to full orange, is exhibited by them. The yellows prevail also in those Noctuæ, the trivial names of which Linné made to end in ago, as N. Fulvago, Citrago, c. I must not conclude this part of my subject without noticing one of the most striking ornaments of the wings of Lepidoptera, the many-coloured eyes which decorate so large a number of them. Some few birds, as the Peacock and Argus Pheasant, have been decked by their Creator very conspicuously with this almost dazzling glory; but in the insects just named it meets us every where. Some, as one of our most beautiful butterflies, Vanessa Io[1970], have them both on the primary and secondary wings; others, as Noctua Bubo[1971], only on the primary; others again, as Smerinthus ocellata[1972], only on the secondary: in some also they are on both sides of the wing, as in Hipparchia Ægeria[1973], and in others only on the upper side, as in Vanessa Io; in others again only on the under side, as in Morpho Teucer[1974]: in some likewise they are very large, as in the secondary wings of the same butterfly: and in others very small, as in those in the wings of the blues (Lycæna). Once more, in some
  • 63. they consist only of iris and pupil, as in Hipparchia Semele, and in others of many concentric circles besides, as in Morpho Teucer, c. v. Legs[1975]. We are next to consider those organs of motion affixed to the trunk, by which insects transport themselves from one place to another on the earth or in the water, and by which also they perform various operations connected with their economy[1976]. In treating of them we should consider their number; kind; substance; articulation with the trunk; position; proportions; clothing; composition; folding; and motions. 1. Number. Having before very fully explained to you the number and kind of the legs of insects in their preparatory states[1977], I shall now confine myself to the consideration of these organs in their perfect or last state; beginning with their number. Insects, properly so called, as I formerly observed[1978], in this state, including the anterior pair or arms, have only six legs, none exceeding or falling short of this number; but in several of the Diurnal Lepidoptera (Vanessa, c.) the anterior pair are spurious, or at least not used as legs, the tarsi having neither joints nor claws[1979]; this in some cases is said to be only a sexual distinction[1980]. In Onitis, Phanæus, and some other Scarabæidæ Mc L., the arm has either none or a spurious tarsus or manus[1981]; which in the first of these genera is also a sexual character. From both these instances we see that walking is only a secondary use of forelegs in the insect tribes. Besides insects proper, a whole tribe of mites (Caris Latr., Leptus Latr., Astoma Latr., Ocypete Leach) have only six legs; the rest, and the Arachnida in general, have eight; in the Myriapods, Pollyxenus has twelve pairs; Scutigera has fifteen; the terrestrial Glomerides (G. Armadillo, c.) sixteen; and the oceanic (G. ovalis) twenty; the oriental Scolopendræ Leach, twenty-one; Polydesmus has usually about thirty pairs; Craspedosoma, fifty; Geophilus electricus at least sixty; in Iulus terrestris there are more than seventy; in I. sabulosus nearly one hundred; in I. fuscus, 124; and in I. maximus 134 pairs or 268 single legs. But with respect to the Geophili, Iuli, c., it is to
  • 64. be observed, that the number of pairs varies in different individuals; and the circumstance that has been before mentioned[1982], that these animals keep acquiring legs in their progress to the perfect state, instead of losing them, renders it difficult to ascertain what is the natural number of pairs in any species. 2. Kinds. Upon a former occasion I gave you a sufficiently full account of the kinds of legs[1982], and I have also assigned my reasons for giving a different denomination to the anterior legs under certain circumstances[1983]; I shall not therefore enlarge further upon this head. 3. Substance. The substance of the legs is generally regulated more or less by that of the rest of the body, only in soft-bodied insects they seem usually more firm and unbending. Each joint is a tube, including the moving muscles, nerves, and air vessels. 4. Articulation with the Trunk. M. Cuvier has observed that the hip (coxa), which is the joint that unites the leg with the body, rather inosculates, in its acetabulum, than articulates in any precise manner[1984]; but this observation, though true of a great many, will not apply universally, for the legs of Orthopterous insects, and of most of the subsequent Orders, are suspended rather than inosculating. Even in many Coleoptera a difference is observable in this respect. I have before mentioned that what are called the puncta ordinaria, which distinguish the sides of the prothorax of many Scarabæidæ and Geotrupidæ, form a base for an elevation of the interior surface with which the extremity of the base of the clavicle, which plunges deep into the breast, ginglymates[1985]; this structure may also be found in other Lamellicorns, as the stag-beetle (Lucanus) and Dynastes, that have not those excavations; in these last it is an elevated ridge forming a segment of a circle with, it should seem, a posterior channel, receiving a corresponding cavity and protuberance of the clavicle. With regard to the mid-leg, in Copris, the coxa is emboxed in a nearly longitudinal cavity of the medipectus, and the coxa of the hind-leg anteriorly is suspended to
  • 65. a transverse cavity of the postpectus, but posteriorly it is received by a cavity of the first segment of the abdomen; so that it may be regarded as suspended anteriorly, and inosculating posteriorly. In some tribes of this Order, as the Weevils (Curculio L.) and Capricorns (Cerambyx), the coxæ of the four anterior legs are subglobose[1986] and extremely lubricous, and are received each by a socket that fits it, and is equally lubricous. In the bottom of this externally, and in the head of the coxa, is an orifice for the transmission of muscles, nerves, and bronchiæ; but the coxa is suspended by ligament in the socket. This structure approaches as near the ball and socket as the nature of the insect skeleton will permit; the high polish of the articulations acts the part of synovia, and the motion is in some degree rotatory or versatile, whereas in Copris, c., lately mentioned, it seems to be more limited, and is probably, at least in the mid- and hind-legs, only in two directions; in the middle pair, probably, from the coxæ being in a position parallel with the breast, opposite to that of the hind pair. In Dytiscus L., Carabus L., and some other beetles, the coxæ, especially the posterior pair, appear to be fixed and incapable of motion. In many insects these coxæ seem to belong as much to the abdomen as to the trunk. We have just seen this to be the case in Copris, c.; and in the Lepidoptera, if the former be separated from the latter, the legs will be detached with it. 4. Location. We are now to consider the location and position of the legs, both in general and with respect to each other. And first, as has been before stated, we may observe that, in the hexapods with wings, the arms belong to the manitrunk, and are attached to the antepectus on each side the prosternum; and the two pair of legs to the alitrunk, the mid-legs being attached to the medipectus, between the scapularia and mesosternum; and the hind-legs to the postpectus, between the parapleura and the posternum; and further, that the arms are opposed to the prothorax: the mid-legs to the mesothorax and the primary organs of flight; and the hind-legs to the metathorax and the secondary organs of flight; though in some
  • 66. cases the wings appear to be behind the legs and in others before them: thus, in Panorpa, the former are nearer the head than the latter; but in the Libellulina the reverse of this takes place, the legs being much nearer the head than the wings: in both cases, however, the scapularia and parapleuræ run from the legs to the wings, but in an oblique direction; and in Panorpa these pieces assume the appearance of articulations of the legs. In most of the apterous hexapods they appear to be attached laterally between the thorax and the pectus[1987]; but in the flea (Pulex) they are ventral. In this tribe the arms are usually stated to be inserted in the head[1988]: but I once succeeded in separating the head of a flea from the trunk, and these organs remained attached to the latter[1989]. As to the Octopods and Arachnidæ, in the mites (Acarus L.) they are lateral, and in their analogues, the spiders (Aranea L.), they emerge between the thorax and the breast, which last they nearly surround; in the Phalangidæ the bases of the coxæ approach near to each other, being separated only by a narrow sternum; in their antagonists, Chelifer and Scorpio, they apply to each other, the anterior ones acting as maxillæ. In the myriapods the legs of the Chilopoda Latr., and some Chilognatha, as Glomeris, are inserted laterally, a single pair in a segment; but in Iulus L. their attachment is ventral, the coxæ seem to spring from a common base, and there are two pair to each segment[1990], except the three first, which bear each a single pair. I shall next consider how the legs are located with respect to each other. To render this clear to you I shall represent each of the variations, which amount in all to twelve in the hexapods that have fallen under my notice, by six dots. 1. In this arrangement the legs are all planted near to each other, there being little or no interval between the pairs, and between the legs of each pair. It is exemplified in the Lepidoptera, Blatta, and many Diptera.
  • 67. 2. Similar to the preceding, but the anterior pair are distant from the two posterior; exemplified in the bees (Apis) and most Hymenoptera; Chironomus; Scutellera; Pachysoma K.[1991] 3. Like the last, but the posterior pair is distant from the two anterior. Examples: Silpha, Necrophorus, Telephorus, c. 4. Similar to the last, but the legs of the posterior pair are more distant from each other than the four anterior. Ex. Curculio L. 5. The legs of each pair near each other, but the pairs distant. Ex. Gibbium. 6. Both the legs of each pair and the pairs distant. Ex. Blaps, c. 7. Anterior pair distant from the two posterior, and the legs of the middle pair rather more distant from each other than those of the other pairs. Ex. Scarabæus Mc L. 8. Like the preceding, only the legs of the middle pair are at a much greater distance from each other. Ex. Copris Mc L. 9. Legs of the two posterior pairs distant. Ex. Hister, Scaphidium. 10. Like the preceding, but the posterior legs more distant than those of the middle pair. Ex. Lygæus. 11. Like the last, but the legs of the anterior pair also distant. Ex. Velia. 12. The arms distant, intermediate legs more distant, posterior legs close together. Ex. Byrrhus L. 5. Proportions. In general the legs of some insects are disproportionally long and slender, as in Phalangium Opilio and some
  • 68. species of Gonyleptes[1992]: those of others are disproportionally short, as in Elater, c. With regard to their relative proportions, the most general rule is, in Hexapods, that the anterior pair shall be the shortest and most slender, and the posterior the longest and thickest; but there are many exceptions: thus, in Macropus longimanus, Clytra longimana, c., in the male the arms are the longest; again, a thing that very rarely occurs, in the same sex of Podalirius retusa the intermediate legs are the longest[1993]; but in Rhina barbirostris and many weevils they are the shortest: in Saperda hirtipes Oliv.[1994] the hind-legs are disproportionally long: with regard to thickness, they are in general extremely slender in Cicindela, and in the Scarabæidæ very thick. In Goliathus Cacicus the arms are more robust than the four legs[1995]; in Gyrinus the latter are more dilated than the former; in many Rutelidæ, and particularly in the celebrated Kanguroo beetle (Scarabæus Macropus Franc.) the hind-legs are much the thickest; in a new genus of weevils from Brazil (Plectropus K.), the intermediate pair are more slender than either the arms or the posterior pair. 6. Clothing. The hairs on the legs of insects, though at first sight they may seem unimportant, in many cases are of great use to them, both in their ordinary avocations and motions: but as most of these were sufficiently noticed when I treated of the sexes of insects[1996], I shall not here repeat my observations, but confine myself to cases not then adverted to. Some insects have all their legs very hairy, as many spiders, the diamond beetle (Entimus imperialis), or at least a species very near it and common in Brazil[1997], c.: in others they are nearly naked, as in the stag- beetle. In the Crepuscular Lepidoptera (Sphinx L.) and some of the Nocturnal ones (Bombyx L.) the thighs are much more hairy than the rest of the legs: and in Lucanus, Geotrupes, and many other Lamellicorns, c., the anterior ones have a yellow or golden spot at their base, composed of decumbent hairs, which prevent them from suffering by the violent friction to which they are exposed in burrowing. In most Petalocerous beetles the tibiæ are set with
  • 69. scattered bristles, and sometimes the thighs. The Tiger beetles (Cicindela) are similarly circumstanced: but the bristles, which are white, are generally arranged in rows. In Dytiscus, Hydrophilus, c., the four posterior tarsi; and in Notonecta the posterior pair, and also the tibiæ—are fringed on each side with a dense series of hairs, which structure assists them in swimming[1998]. The tarsi, especially the anterior pair, in a certain family of Lamia F. (L. papulosa, c. [1999]), are similarly fringed, only the hairs curl inwards; and the hand in Sphex and Ammophila, but not in Pelopæus and Chlorion, is fringed externally with long bristles. 7. Composition. With regard to their composition, both arms and legs generally consist of five pieces, which Entomologists have denominated—the coxa or hip—the trochanter—the femur or thigh— the tibia or shank—and the tarsus or foot. Where the structure and use of the fore-leg is different from that of the four hind-legs, I propose calling these pieces by names corresponding with those which anatomists have appropriated to the arm in the higher vertebrate animals: thus, as you will see in the table, I call the whole fore-leg the brachium or arm; and the coxa becomes the clavicula or collar-bone; the trochanter, the scapula or shoulder-blade; the femur, the humerus or shoulder; the tibia, the cubitus or arm; the tarsus, the manus or hand. But let me not lead you to suppose that the pieces, either in the arms or legs of insects, which are there named after certain others in vertebrate animals, precisely correspond with them—by no means—since that is a very doubtful point; and some of them, as the trochanter, clearly do not. Many gentlemen skilled in anatomy, as I have before observed[2000], have thought that what is regarded as the coxa in insects really represents the femur: but there are considerable difficulties in the way of this supposition, several of which I then stated. I shall not however enter further into the subject, and take the above names; since this application of them is so general and so well understood, except with regard to the fore-leg, under certain circumstances, as I
  • 70. find them. I shall now consider them in the order in which I have named them. a. Coxa or Clavicula[2001]. The coxa is the joint that connects the leg with the trunk of the insect. With regard to their shape, the most general form of the four anterior is more or less that of a truncated cone: in the Staphylinidæ, however, they tend to a pyramidal or four-sided figure; as do the whole six in the Trichoptera: in numbers of the weevils and capricorns they are subglobose; in the Lamellicorns they are mostly oblong, and not prominent: the posterior pair in the Coleoptera are generally flat and placed in a transverse position, and more or less oblong and quadrangular: in Elater, c., they are cuneiform: in Haliplus Latr. they are dilated, and cover the thigh[2002]: in Buprestis, Copris, c., they have a cavity that partly receives it: the corresponding part, the clavicle, in the arm of Gryllotalpa, is very large and remarkable; viewed underneath it is triangular, and trifid where the trochanter articulates with it: in that of Megachile Willughbiella the clavicle is armed with a spine[2003]. As to their proportions, the most general law seems to be, that the anterior pair shall be the shortest and smallest, and the posterior the longest and largest. In some instances, as in Buprestis, the two anterior pair are nearly equal; in others (Mantis, Eurhinus K.), the anterior are the longest, in the former as long as the thigh, and the four posterior the shortest: in the Trichoptera, Lepidoptera, c., all are nearly equal; in Mantis the two posterior, and in Phengodes the intermediate pair are the largest; but in Necrophorus they are the smallest:—though almost universally without articulations, in Galeodes the clavicle consists of two and the coxa of three[2004]. b. Trochanter or Scapula[2005]. This is the second joint of the leg: and if the coxa is regarded as the analogue of the thigh in vertebrate animals, this should seem to represent the patella or rotula, vulgarly called the knee-pan. Latreille and Dr. Virey consider this articulation as merely a joint of the coxa[2006]; but if closely examined,
  • 71. especially in Coleopterous insects, you will find it so fixed to the thigh as scarcely to have separate motion from it, and in many cases it seems to be merely its fulcrum; but I am not aware that any instance occurs in which it has not motion separate from that of the former joint. As to its articulation with the coxa,—in the Coleoptera it appears to be of a mixed kind; for it inosculates in that joint, is suspended by ligament to its orifice, and its protuberances are received by corresponding cavities in it; and its cavities receive protuberances, which belongs to a ginglymous articulation. I have observed two variations in this Order, in one of which the motion of the thigh and trochanter is only in two directions, and in the other it is nearly versatile or rotatory. The Lamellicorns afford an example of the first, and the Rhyncophorous beetles or weevils of the second. If you extract from the coxa the thigh with the trochanter of the larger species of Dynastes Mc L., you will find that the head of the latter is divided into two obtuse incurving lobes or condyles: that on the inner side being the smallest and shortest, and constricted just below its apex: and that under this is a shallow or glenoid cavity, terminating posteriorly in a lubricous flat curvilinear ridge. If you next examine the trochanter in articulation with the coxa, you will perceive that the head of the former inosculates in it, that the lower condyle is received by a sinus of the coxa, which also has a lubricous very shallow cavity corresponding with the ridge, in which it turns; and in the head of the coxa, on the lower side, is an external condyle, which is received by a sinus common to both, of the head of the thigh and of the exterior side of the trochanter[2007], in which it likewise turns: this last condyle has also an internal protuberance, which appears to ginglymate with a cavity of the trochanter: from this structure the leg is limited chiefly to a motion up and down upon two pivots, or to fold and extend itself. You will find an articulation very near this, but on a smaller scale, in the stag-beetle. In the other kind of articulation, which admits of freer motion, the head of the trochanter is prolonged, and the process terminates in a short interior condyle, which appears to work in a corresponding cavity of
  • 72. the interior of the coxa; and the base of the process is encompassed by a ridge with a cavity behind it, which is received by another of the lower part of that piece, and admits a corresponding ridge—a structure that allows a rotatory motion. In the hind-legs of this tribe the motion is chiefly limited to folding and extending; in Carabus, c., also the head of the trochanter is nearly hemispherical, and the articulation approaches ball and socket. In most of the other Orders, the Hymenoptera excepted, there is little or no inosculation, the trochanter being simply suspended by ligament to the coxa as well as to the thigh; its connection with the latter is similar in Coleoptera; but in Cicindela, c., it inosculates in it. The part we are considering varies in its position with respect to the thigh: in the hind-legs of Carabus, c., it forms a lateral fulcrum on the inner side of that part, and does not intervene between its base and the coxa; the muscles from the latter entering the former, not at the bottom of the base, but at its side: but in the four anterior legs it forms their base, as it does in all the legs in Apion, and in all the Orders except the Coleoptera, cutting them entirely off from contact with the coxa: in the Lamellicorns they cut off part of the base obliquely, but so as to permit their coming in contact with the condyle of the coxa, as before mentioned. In the Ichneumonidæ and some other Hymenoptera the trochanter appears to consist of two joints particularly visible in the posterior legs[2008]. As to size in general,—the part in question is smaller than the coxa; but in Notonecta it is larger, and in the dog-tick (Ixodes Ricinus) longer than that joint. It exhibits few variations in its shape or appendages worthy of particular notice. In general, in the Coleoptera it is triangular or trigonal; but in Carabus L., in the hind-leg it is oblong or rather kidney-shaped; in that of Necrophorus[2009] it terminates in one or two teeth or spines, varying in length in the different species: in the other Orders it is not remarkable in this respect. c. Femur or Humerus[2010]. The femur or thigh is the third, and usually the largest and most conspicuous joint of the leg. In the
  • 73. hypothesis before alluded to[2011] it is considered as the analogue of the tibia of vertebrate animals. With regard to the articulation of this part with the trochanter, it has been sufficiently explained under that head, and that with the tibia I shall treat of when I come to that joint. As to the size of the thighs, and their relative proportions to each other and to the remaining joints of the leg, the most general law is, that the anterior pair shall be the shortest and smallest, and the posterior the longest and largest. With respect to the remaining articulations, most commonly the thigh is longer and larger than the tibia, and the tibia than the tarsus. But there are numerous exceptions to both these rules. With respect to the first, we may begin by observing that the increase of the magnitude of the thigh, from the anterior to the posterior pair, is usually gradual: but in many jumping insects, and likewise many that do not jump, the posterior pair are suddenly and disproportionally thicker than the rest[2012]. Again, in many insects the anterior pair are the longest and thickest, as in Macropus longimanus, Bibio, Nabis, c.: in others, the intermediate exceed the rest in magnitude, as in Onitis Aygulus, cupreus; Sicus flavipes, c.; in many Lamellicorns all the thighs are incrassated and nearly equal in size: but in some, as Ryssonotus nebulosus Mc L.[2013], the intermediate pair are rather smaller than the rest. With respect to the second rule—in some, as in the male of Macropus longimanus, the anterior tibia, though more slender, is longer than the thigh; in Hololepta maxillosa it is longer and more dilated; in Lamia marmorata, or one related to it from Brazil, the intermediate pair are longer; in Ateuchus gibbus and others of that tribe the posterior thighs are smaller than the tibiæ: and, to mention no more; in Callichroma latipes the posterior tibia is wider than the part last named. Again, the tarsi are as long as either tibia or thigh in many of the larger Dynastidæ, as Megasoma Actæon, c.; longer than either in Melolontha subspinosa F.; and in Tiphia, Scolia and affinities, often as long, or longer than both together.
  • 74. As to shape,—the thigh, especially in the fore-leg, varies considerably: most generally it is flat, linear, and a little thicker where it is united to the tibia, on the outer side convex, and concave next the body; but in many it is gradually thicker from the base to the apex: in some Cerambyces (C. thoracicus) it is clavate; in others of this genus and Molorchus they may be called capitate; in Pterostichus they are rather lanceolate; in Onitis Sphinx the humerus is triangular, and the intermediate thigh rhomboidal; in Bruchus Bactris it is bent like a bow; and in some Brazilian Halticæ it is nearly semicircular. The humerus in Phasma is attenuated at the base; in Empusa gongyloides it is at first ovato-lanceolate, and terminates below in a kind of footstalk[2014]; in Phasma flabelliforme it is dolabriform[2015]; in Mantis often semioval or semielliptical, and thickest at the inner edge, which affords space for two rows of spines with which it is planted. In Phyllium siccifolium all the thighs are furnished on both sides with a foliaceous appendage nearly from base to apex[2016]: in a species of Empusa (E. macroptera), the four posterior ones are so distinguished only on their posterior side[2017]: others of this last genus, as E. gongyloides, have an alary appendage on both sides at the apex of these thighs[2018]; and another family, as E. pauperata, have only one on the posterior side[2019]. The thighs of no insect are more remarkable for their elegant shape,—tapering gradually from the base to the apex, where they swell again into a kind of knee,—than the posterior ones of the locusts (Locusta Leach); each side of these thighs is strengthened with three longitudinal nearly parallel ridges, and the upper and under sides are adorned by a double series, in some coalescing as they approach the tibia, of oblique quadrangular elevations resembling scales[2020]. I shall next say a few words upon the spines and other processes which arm the thigh. Those moveable ones of Mantis which help to form a fearful instrument of destruction, have just been mentioned, and similar ones, but less conspicuous, arm the intermediate thighs of Sicus flavipes: other appendages of this kind are for a less
  • 75. destructive purpose—to keep the tibia when folded in its place. This seems to be the use of the serratures and spine that arm the thigh of Bruchus Bactris, or the Hymenopterous genera Leucospis, Chalcis, c.; in Onitis Aygulus a short filiform horn arms the humerus, and a longer crooked one that of many species of Scaurus[2021]. In many Stenocori the thighs terminate in two spines, and in Gonyleptes K. the posterior ones are armed internally with very strong ones; with which, as the legs converge at their knee[2022], they may probably detain their prey. The knee-pan (Gonytheca) of the thigh, or the cavity at its end, which receives the head of the tibia, is very conspicuous in the weevils; but in no insects more than in Locusta[2023], in which tribe it deserves your particular attention. d. Tibia or Cubitus[2024]. The tibia or shank is the fourth joint of the leg, which according to the hypothesis lately alluded to is the analogue, in the anterior leg of the carpus or carpal bones, and in the four posterior ones of the tarsus or tarsal bones of vertebrate animals. This may be called the most conspicuous of the articulations of the leg; for though it is generally more slender and often shorter than the thigh, it falls more under the eye of the observer, that joint being more or less concealed by the body: it consists in general of a single joint; but in the Araneidæ and Phalangidæ it has an accessory one, often incrassated at its base, which I have named the Epicnemis[2025]. With respect to the articulation of the tibia with the thigh—we may observe that in general it is by means of three processes or condyles, two lateral and one intermediate, of the head of the former joint[2026]: the lateral ones are usually received by a cavity or sinus of the gonytheca of the thigh[2027]; and upon these the tibia turns, with a semirotatory motion up and down as upon a pair of pivots: at the same time the mola or head of the latter joint, which has often a flexure so as to form an elbow with the rest of it, inosculates in the gonytheca, and is also suspended by ligament to the orifice through which the muscles, nerves, and bronchiæ are
  • 76. transmitted: so that in fact the articulation, strictly speaking, belongs exclusively to none of the kinds observable in vertebrate animals, but partakes of several, and may properly be denominated a mixed articulation,—a term applicable in numerous instances also to the other articulations of the legs of insects. In the different Orders some variations in this respect take place,—I will notice some of the most remarkable. In no Coleopterous insects is the structure more distinctly visible than in the larger Lamellicorns. In Copris bucephalus, for instance, if you divide the thigh longitudinally, you will find on each side, at the head, that it is furnished with a nearly hemispherical protuberance, perforated in the centre for the transmission of muscles, and surrounded externally by a ridge, leaving a semicircular cavity between them[2028]: if you next examine the tibia, after having extracted it, you will find on each side, at the base, a cavity corresponding with the protuberance of the thigh which it receives, having likewise a central orifice, and surrounded by a semicircular ridge corresponding with the cavity in the thigh in which it acts: below this ridge another cavity, forming a small segment of a circle, receives the ridge of the thigh[2029]. You will observe that the ridge of the tibia represents the lateral condyle lately noticed: in the Dynastidæ this is more prominent, and often forms a smaller segment of a circle. In these also the protuberance of the thigh is more minute, and its ridge is received by a cavity of the tibia nearly semicircular[2030]; in Geotrupes Latr. the articulation is not very different, though on a reduced scale; in Calandra Palmarum the lateral condyles of the tibiæ are flatter and broader[2031]; and the articulation not being quite so complex, this joint is kept steady by an intermediate process observable in the gonytheca[2032]. From the above description it appears that the dislocation of the tibia is effectually prevented in the Lamellicorns by the protuberance and ridge of the thigh working in their corresponding cavities, while the condyle of that part turns with a rotatory motion in the cavity of the thigh. In the Orthoptera Order the tibia is suspended by a ligament, in the gonytheca the lateral condyles, which are very prominent, working in a sinus of that
  • 77. part[2033]. The subsequent Orders exhibit no very striking variations from these types of articulation, I shall therefore not detain you longer upon this head. With regard to the proportions and magnitude of the joint we are considering,—the most general law is, that the anterior pair should be shorter and more slender than the intermediate; and the intermediate than the posterior; and that all the tibiæ should be shorter and more slender than the thighs, and longer and thicker than the tarsi. Various exceptions, however, to this rule in all these cases might be produced; but I shall only observe that in all those insects in which the fore-legs are calculated for digging or seizing their prey, as in the Petalocerous beetles, the Gryllotalpa, Mantis, c., this joint of the leg is usually much enlarged and more conspicuous than the others. As to its figure and shape—most commonly the tibia grows thicker from the base to the apex, as in the majority of Coleoptera, Hymenoptera, c.; in the Orthoptera, Neuroptera, c., it is generally equally thick every where. Another peculiarity relating to this head observable in it, is its tendency to a trigonal figure: this, however, though very general, is not universal;—thus, in some Orthoptera, as Pterophylla K., its horizontal section is quadrangular; in others, as Locusta Leach and many other insects, it is nearly a circle; in some scorpions it is almost a hexagon. The superficial shape also of this joint in numerous instances is more or less triangular, but it sometimes recedes from this form:—thus, in Callichroma latipes it is a segment of a circle; in some Empides it is clavate; in Onitis Sphinx, dolabriform; in the Orthoptera, Neuroptera, c., it is usually linear; in some Lygæi it is angular[2034]: but the most remarkable tibiæ in this respect are those of such species of this last genus as have the posterior ones winged or foliaceous, so that they resemble the leaf of some plant—the tibia being the rachis, and the wing (which in some species is veined) representing the leaf itself. This structure is exemplified in Lygæus compressipes, phyllopus, foliaceus, c.[2035] Under this head I must say a few words upon the flexure of this
  • 78. Welcome to our website – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookbell.com