Robot Systems for Rail Transit Applications 1st Edition Hui Liu

Robot Systems for Rail Transit Applications 1st
Edition Hui Liu download
https://textbookfull.com/product/robot-systems-for-rail-transit-
applications-1st-edition-hui-liu/
Download more ebook from https://textbookfull.com

We believe these products will be a great fit for you. Click
the link to download now, or visit textbookfull.com
to discover even more!
Designing Robot Behavior in Human Robot Interactions
1st Edition Changliu Liu (Author)
https://textbookfull.com/product/designing-robot-behavior-in-
human-robot-interactions-1st-edition-changliu-liu-author/
Rail Transport Systems Approach 1st Edition Aleksander
S■adkowski (Eds.)
https://textbookfull.com/product/rail-transport-systems-
approach-1st-edition-aleksander-sladkowski-eds/
Smart Device Recognition: Ubiquitous Electric Internet
of Things Hui Liu
https://textbookfull.com/product/smart-device-recognition-
ubiquitous-electric-internet-of-things-hui-liu/
Conjugated Polymers for Biological and Biomedical
Applications 1st Edition Bin Liu
https://textbookfull.com/product/conjugated-polymers-for-
biological-and-biomedical-applications-1st-edition-bin-liu/

Trends in Control and Decision Making for Human Robot
Collaboration Systems 1st Edition Yue Wang
https://textbookfull.com/product/trends-in-control-and-decision-
making-for-human-robot-collaboration-systems-1st-edition-yue-
wang/
Noise and Vibration Mitigation for Rail Transportation
Systems Proceedings of the 14th International Workshop
1st Edition Xiaozhen Sheng
https://textbookfull.com/product/noise-and-vibration-mitigation-
for-rail-transportation-systems-proceedings-of-the-14th-
international-workshop-1st-edition-xiaozhen-sheng/
Robot Intelligence Technology and Applications 2
Results from the 2nd International Conference on Robot
Intelligence Technology and Applications 1st Edition
Jong-Hwan Kim
https://textbookfull.com/product/robot-intelligence-technology-
and-applications-2-results-from-the-2nd-international-conference-
on-robot-intelligence-technology-and-applications-1st-edition-
jong-hwan-kim/
Robot Intelligence Technology and Applications 3
Results from the 3rd International Conference on Robot
Intelligence Technology and Applications 1st Edition
Jong-Hwan Kim
https://textbookfull.com/product/robot-intelligence-technology-
and-applications-3-results-from-the-3rd-international-conference-
on-robot-intelligence-technology-and-applications-1st-edition-
jong-hwan-kim/
Sliding Mode Control Methodology in the Applications of
Industrial Power Systems Jianxing Liu
https://textbookfull.com/product/sliding-mode-control-
methodology-in-the-applications-of-industrial-power-systems-
jianxing-liu/

Robot Systems for Rail Transit
Applications
Hui Liu
School of Traffic and
Transportation Engineering,
Central South University,
Changsha, China

Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
Copyright © 2020 Elsevier Inc. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including photocopying, recording, or any information storage and retrieval system, without permission
in writing from the publisher. Details on how to seek permission, further information about the Publisher’s
permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the
Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the Publisher (other than
as may be noted herein).
Notices
Knowledge and best practice in this ﬁeld are constantly changing. As new research and experience broaden our
understanding, changes in research methods, professional practices, or medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any
information, methods, compounds, or experiments described herein. In using such information or methods they
should be mindful of their own safety and the safety of others, including parties for whom they have a professional
responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability
for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or
from any use or operation of any methods, products, instructions, or ideas contained in the material herein.
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
ISBN: 978-0-12-822968-2
For information on all Elsevier publications visit our website at
https://www.elsevier.com/books-and-journals
Publisher: Matthew Deans
Acquisitions Editor: Glyn Jones
Editorial Project Manager: Naomi Robertson
Production Project Manager: Nirmala Arumugam
Cover Designer: Matthew Limbert
Typeset by TNQ Technologies

List of figures and tables
Figure 1.1 Robots in the manufacturing, dispatch, and maintenance of rail transit.
Figure 1.2 Role of robots in rail transit maintenance.
Figure 1.3 Key problems of rail transit robot systems.
Figure 2.1 Working steps of the assembly robot.
Figure 2.2 Different types of assembly robots.
Figure 2.3 Overall frame diagram of rail transit assembly robot system.
Figure 2.4 Main components of assembly robot.
Figure 2.5 Mechanical diagram of assembly robot: (A) base component, (B) rotating
joint, (C) arm connecting component, (D) wrist joint, and (E) end
effector.
Figure 2.6 Trajectory planning algorithms.
Figure 2.7 Process of using the artificial neural network (ANN) for inverse dynamics
calculation.
Figure 3.1 Multirobot collaboration.
Figure 3.2 Humanerobot collaboration.
Figure 3.3 The block diagram of collaborative robot system.
Figure 3.4 Sensors for collaborative robots.
Figure 3.5 Classification of end effectors.
Figure 3.6 Feature extraction algorithms.
Figure 3.7 The flow chart of the HOG algorithm.
Figure 3.8 The flow chart of the SIFT algorithm.
Figure 3.9 The flow chart of the LBP algorithm.
Figure 3.10 Target detection algorithms.
Figure 3.11 Target tracking algorithms.
Figure 4.1 Main content diagram of automatic guided vehicles (AGVs).
Figure 4.2 Navigation methods. AGV, automatic guided vehicle; LiDAR, light detec-
tion and ranging; SLAM, simultaneous localization and mapping.
Figure 4.3 Global path planning algorithms.
Figure 4.4 Local path planning algorithms.
Figure 4.5 Humanerobot interaction algorithms.
Figure 4.6 Flowchart of the hybrid path planning model. KELM, kernel-based extreme
learning machine; QPSO, quantum particle swarm optimization.
Figure 5.1 The advantages of Autonomous rail Rapid Transit (ART).
Figure 5.2 The overview diagram of Autonomous rail Rapid Transit (ART).
ix

Figure 5.3 Schematic diagram of ART sensor fusion.
Figure 5.4 The core of the pedestrian detection algorithm.
Figure 5.5 The flow chart of Histogram of Oriented Gradient (HOG) feature þ
Support Vector Machine (SVM) classification.
Figure 5.6 The flow chart of the Support Vector Machine (SVM) classification.
Figure 5.7 The flow chart of pedestrian contour extraction.
Figure 5.8 Posture recognition process.
Figure 6.1 Inspection robot structure.
Figure 6.2 The railway applications of the inspection robots.
Figure 6.3 Main components of the inspection robots.
Figure 6.4 Rail transit inspection robot technologies.
Figure 7.1 Advantages of dual-arm robots.
Figure 7.2 Channel robots Trouble of Moving Electric Multiple Units Detection
System diagram.
Figure 7.3 Structure of ground track.
Figure 7.4 Forming diagram of infrared image formation.
Figure 7.5 Analysis process of visible image. HOG, histogram of orientation gradient.
Figure 7.6 Intelligent analysis method of infrared thermal image. ANFIS, adaptive
network-based fuzzy inference system; BP, backpropagation.
Figure 7.7 Bogie fault diagnosis structure diagram.
Figure 7.8 Flow of fault diagnosis. EMD, empirical mode decomposition; WPT,
wavelet packet transform.
Figure 7.9 The sigmoid function.
Figure 7.10 The tanh function.
Figure 7.11 The rectified linear unit function.
Figure 8.1 Main components of rail transit inspection unmanned aerial vehicles
(UAVs). GPS, global positioning system; IMU, inertial measurement unit.
Figure 8.2 Scheduling system of an unmanned aerial vehicle (UAV).
Figure 8.3 Flowchart of perception algorithm. ROI, region of interest.
(A) The real-time image stabilization of the collected video is finalized.
(B) The track region of interest is extracted from the perception image
according to railway boundary regulations, to reduce unnecessary
operations and speed the subsequent image processing rate.
(C) The intrusion is identified based on a track map.
Table 2.1 Advantages and disadvantages of three assembly methods.
Table 2.2 Advantages and disadvantages of joint space trajectory planning and Carte-
sian space trajectory planning.
Table 3.1 Comparison of traditional industrial robots and collaborative robots.
Table 3.2 Several mainstream collaborative robots and manufacturers.
Table 5.1 Comparison of main technical solutions of ART and modern tram signal
system.
Table 5.2 Traditional pedestrian detection algorithms.
x

Table 5.3 Pedestrian detection algorithms based on deep learning.
Table 5.4 The pedestrian posture recognition accuracy of BP neural network.
Table 7.1 Composition of laser sensor.
Table 7.2 Functions and characteristics of two-dimensional laser sensor.
Table 7.3 Comparison of channel robot wireless recharging methods.
xi

Preface
Rail transit is the lifeblood of many national economies and the backbone of transportation.
Safe and efficient rail transit is based on highly reliable manufacturing and high-quality
maintenance. Rail transit robots can replace manual repetitive tasks and improve the auto-
mation level of the rail transit system. As a result, work efficiency can be improved, acci-
dental failures due to human negligence can be avoided, and the safety of the rail transit
system can be improved.
Rail transit robots involve the intersection of automatic control, artificial intelligence, signal
processing, pattern recognition, mechanical engineering, and transportation engineering.
When applied to the rail transit system, the robot is faced with the key problems to be
solved urgently. Therefore, rail transit robots are currently recognized as research hotspots
among scientific problems. Based on research from the past 10 years, the author puts
forward a framework of rail transit robot technology and completes the related work.
This book covers seven mainstream rail transit robots, including assembly robots,
collaborative robots, automated guided vehicles, autonomous rail rapid transit, inspection
robots, channel robots, and inspection unmanned aerial vehicles. The key problems of
robots are described in detail, including positioning navigation, path planning,
humanerobot interaction, and power management, etc. For students and managers in related
departments, this book can provide valuable information about rail transit robots. For
researchers and doctoral students, this book can provide some ideas and encourage future
research in rail transit robots.
This book contains eight chapters:
Chapter 1: Introduction
This chapter first outlines the rail transit robot. Then the chapter describes three basic issues
of robotics, including navigation, humanerobot interaction, and power control.
Chapter 2: Rail transit assembly robot systems
This chapter first introduces the development progress and key technologies of assembly
robots, and then introduces the main components of assembly robots. After that, the
dynamics models for the assembly robot systems are explained. Finally, the artificial neural
network algorithm for the inverse dynamic optimization calculation of the robot arm is
introduced.
xiii

Chapter 3: Rail transit collaborative robot systems
This chapter first gives the basic definition of a collaborative robot. Then the chapter
summarizes the development history, application field and collaborative mode of
collaborative robots. The components of the collaborative robot are summarized. Finally, the
basic concepts of visual perception in human-robot collaboration are introduced.
Chapter 4: Automatic Guided Vehicles (AGVs) in the rail transit intelligent
manufacturing environment
This chapter first introduces the development progress and types of the AGV in rail transit
intelligent manufacturing environment. Then the main components of AGVs are introduced.
After that, the key technologies and the applications of the AGV are introduced.
Chapter 5: Autonomous rail Rapid Transit (ART)
This chapter firstly introduces the hardware of ART. Then, the technologies of ART are
introduced. Finally, the pedestrian detection algorithms of ART are introduced in detail.
Chapter 6: Rail transit inspection robots
This chapter first introduces the development history, function and main components of the
rail transit inspection robot. Then, two key technologies to ensure that the inspection robots
normally complete the inspection work, positioning methods, path planning methods, and
handeeye vision system are introduced in detail.
Chapter 7: Rail transit channel robot systems
This chapter firstly introduces the development history and main components of the rail
transit channel robot, including the ground rail, dual-arm robot, infrared thermometer, laser
sensor, etc. Then the TEDS intelligent sensing system is described in detail. Finally, fault
diagnosis algorithms based on deep learning models are introduced.
Chapter 8: Rail transit inspection Unmanned Aerial Vehicle (UAV)
This chapter first introduces the development history of the UAV and its applications in
various fields. Secondly, it introduces the basic structure of fixed-wing UAVs, unmanned
helicopters and rotary-wing UAVs. Various sensors are applied to rail transit inspection.
Then the UAV technologies are described in detail. Finally, the applications of the UAV in
the detection of rail transit intruding detection are introduced.
Prof. Dr.-Ing. habil. Hui Liu
Changsha, China
November 2019
Preface
xiv

Acknowledgement
The studies in the book are supported by the National Natural Science Foundation of China,
the National Key R&D Program of China, and the Innovation Drive of Central South
University, China. The publication of the book is funded by the High-level Postgraduate
Text Book Project of the Hunan Province of China. In the process of writing the book,
Mr. Zhu Duan, Mr. Jiahao Huang, Mr. Kairong Jin, Mr. Yu Xia, Mr. Rui Yang, Ms. Shi Yin,
Mr. Ye Li, Mr. Guangji Zheng, Ms. Jing Tan, Mr. Huipeng Shi, Mr. Haiping Wu, Mr. Chao
Chen, Mr. Zhihao Long, and other team members have done a lot of model verification and
other work. These team members as mentioned have the same contribution to this book.
xv

Nomenclature list
#
2D Two-Dimensional
3C Computer, Communication, Consumer Electronic
3D Three-Dimensional
A
ABB Asea Brown Boveri
AC Alternating Current
ACMS Aircraft Condition Monitoring System
ACO Ant Colony Optimization
ADC Analog-to-Digital Converter
ADU Automatic Drilling Unit
AGV Automatic Guided Vehicle
AGVS Automated Guided Vehicle System
AMR Anisotropic Magnetoresistive
ANN Artificial Neural Network
ANIFS Adaptive Network-Based Fuzzy Inference System
AR Augmented Reality
ARM Advanced RISC Machine
AP Access Point
APF Artificial Potential Field
API Application Programming Interface
ARMA Autoregressive Moving Average
ART Autonomous rail Rapid Transit
ASK Amplitude Shift Keying
ATC Automatic Train Control
ATO Automatic Train Operation
ATP Automatic Train Protection
ATS Automatic Train Supervision
AUC Area Under Curve
xvii

B
BFS Best-First Search
BP Back Propagation
BPNN Back Propagation Neural Network
BRIEF Binary Robust Independent Elementary Features
C
CAD Computer-Aided Design
CCD Charge Coupled Device
CCOT Continuous Convolution Operators Tracker
CIMS Computer Integrated Manufacturing System
CNC Computerized Numerical Control
CNN Convolutional Neural Network
CNR China Northern Locomotive Rolling Stock Industry Group
CPU Central Processing Unit
CRF Conditional Random Fields
CSM Correlation Scan Match
CSR China Southern Locomotive Rolling Stock Industry Group
CW Continuous Wave
CWT Continue Wavelet Transform
D
DC Direct Current
D-DCOP Dynamic Distributed Constraint Optimization Problem
Dec-MDP Decentralized Markov Decision Process
DEM Digital Elevation Model
DFT Discrete Fourier Transform
D-H Denavit-Hartenberg
DLT Deep Learning Tracker
DMPC Distributed Model Predictive Control
DNFO Dynamic Network Flow Optimization
DNN Deep Neural Network
DOF Degree of Freedom
DSP Digital Signal Processor
DTW Dynamic Time Warping
E
ECO Efficient Convolution Operator
EKF Extended Kalman Filter
EKF-SLAM Extended Kalman Filter SLAM
ELM Extreme Learning Machine
EMD Empirical Mode Decomposition
EMU Electric Multiple Units
Nomenclature list
xviii

F
FAS Flexible Assembly System
FAST Features from Accelerated Segment Test
Faster RCNN Faster Regional Convolutional Neural Network
FDD Frequency Division Duplexing
FFT Fast Fourier Transform
FMS Flexible Manufacturing System
FSK Frequency Shift Keying
FTP File Transfer Protocol
G
GA Genetic Algorithm
GOA Grade of Automation
GPRS General Packet Radio Service
GPS Global Positioning System
H
HDT Hedged Deep Tracking
HDFS Hadoop Distributed File System
HMM Hidden Markov Model
HOG Histogram of Oriented Gradient
HRI Human-Robot Interaction
HSB Hue-Saturation-Brightness
I
ID Identity Document
IDIM-LS Inverse Dynamic Identification Model and Linear Least Squares Technique
IFF Identification Friend or Foe
IFR International Federation of Robots
IGBT Insulated Gate Bipolar Translator
IL Imitation Learning
IMU Inertial Measurement Unit
IMW Intelligent Manufacturing Workshop
IMF Intrinsic Mode Function
IoU Intersection over Union
INS Inertial Navigation System
ISM Industrial Scientific Medical
J
JFET Junction Field-Effect Transistor
K
KCF Kernelized Correlation Filters
KF Kalman Filter
KNN K-Nearest Neighbors
Nomenclature list
xix

L
LAN Local Area Network
LBP Local Binary Pattern
LCD Liquid Crystal Display
LiDAR Light Detection and Ranging
LRR Long-Range Radar
M
MAE Mean Absolute Error
MANET Mobile Ad Hoc Network
mAP mean Average Precision
MBTA Massachusetts Bay Transit Authority
MDPs Markov Decision Processes
MEEM Multiple Experts using Entropy Minimization
MEMS Micro-Electro-Mechanical Systems
MF Morphological Filter
MIL Multiple Instance Learning
MILP Mixed Integer Linear Programming
MLP Multilayer Perceptron
MSE Mean Square Error
MTSP Multiple Traveling Salesman Problem
MVB Multifunction Vehicle Bus
N
NFS Network File Systems
NMS Nonmaximun Suppression
NOMA Nonorthogonal Multiple Access
NP Nondeterministic Polynomial
O
OFDM Orthogonal Frequency Division Multiplexing
OGAePSO Optimum Genetic AlgorithmeParticle Swarm Optimization algorithm
ORB Oriented FAST and Rotated BRIEF
P
PG Policy Gradient
PLC Programmable Logic Controller
POS Point of Sale
PRM Probabilistic Road Map
PSK Phase Shift Keying
PSO Particle Swarm Optimization
PTZ Pan-Tilt-Zoom
PUMA Programmable Universal Machine for Assembly
Nomenclature list
xx

Q
QPSO Quantum Particle Swarm Optimization
QR Quick Response
R
RANSAC Random Sample Consensus
RBF Radial Basis Function
RBPF Rao-Blackwellized Particle Filter
RCC Remote Center Compliance
RCNN Regional Convolutional Neural Network
ReCNN Regions with CNN features
RDD Resilient Distributed Dataset
ReFCN Region-based Fully Convolutional Networks
RFID Radio Frequency Identification
RGB Red-Green-Blue
RGB-D Red-Green-Blue-Deep
RL Reinforcement Learning
RNN Recurrent Neural Network
ROC Receiver Operating Characteristic curve
ROI Region of Interest
ROS Robot Operating System
RPN Region Proposal Networks
RRT Rapid-Exploration Random Tree
RSSI Received Signal Strength Indication
RTP Real-Time Protocols
S
SARSA State Action Reward State Action
SCARA Selective Compliant Assembly Robot Arm
SDA Stacked Denoising Autoencoder
SEA Series Elastic Actuator
SfM Structure from Motion
SIA Swarm Intelligence Algorithm
SIFT Scale Invariant Feature Transform
SLAM Simultaneous Localization and Mapping
SMT Surface Mount Technology
SPP Spatial Pyramid Pooling
SRDCF Spatially Regularized Discriminative Correlation Filters
SRR Short-Range Radar
SSD Single Shot multibox Detector
SURF Speeded Up Robust Features
SVM Support Vector Machine
SMT Surface Mount Technology
Nomenclature list
xxi

T
TCN Train Communication Network
TCP Transmission Control Protocol
TCSN Train Control and Service Network
TDD Time Division Duplexing
TEDS Trouble of moving EMU Detection System
TOF Time of Flight
U
UAV Unmanned Aerial Vehicle
UDP User Datagram Protocol
UHV Ultrahigh Voltage
UNECE United Nations Economic Commission for Europe
USB Universal Serial Bus
UWB Ultrawideband
V
VR Virtual Reality
VRP Vehicle Routing Problem
VSA Variable Stiffness Actuator
W
WiFi Wireless Fidelity
WPT Wavelet Packet Transform
WTB Wire Train Bus
Y
YOLO You Only Look Once
Nomenclature list
xxii

CHAPTER 1
Introduction
1.1 Overview of rail transit robots
Railway transit is vital to the national economy. Research in Japan and China indicates
that development of the railway can lead to economic growth in the railway field [1,2].
To stimulate economic development, governments are actively developing the railway
transport industry. In such a large industrial system, the development of automation can
improve efficiency and reduce costs, so the level of automation in railways should be
increased. The increase in automation requirements in the rail transit system generates a
pursuit for rail transit robot systems.
In worldwide, many countries have proposed similar strategic plans to encourage the
development of robots in the railway transit system. Taking China as an example, “A
Country With a Strong Transportation Network” and “Smart Railway” are two important
plans that encourage the development of automation of the rail transit system. Driven by
these plans, many Chinese rail equipment manufacturers and operators have carried out
extensive research in robotics. Zhuzhou CRRC Times Electric Co., Ltd. applied an
automatic generating line to produce high-speed train converters [3]. CRRC Qishuyan
Institute Co., Ltd. developed an intelligent manufacturing workshop for gear transmission
systems for high-speed trains. CRRC Zhuzhou Institute Co., Ltd. combined automatic
guided vehicle (AGV) technology with urban transportation equipment and designed
autonomous rail rapid transit (ART) [4].
A variety of robot systems are employed. The use of robot systems in the rail transit
system can be divided into three aspects: manufacturing, dispatch, and maintenance. In
these three parts of rail transit, different kinds of robots have completely different roles, as
shown in Fig. 1.1.
1.1.1 Rail transit robots in manufacturing
“A Country With a Strong Transportation Network” points out that intelligent manufacturing
is required for the rail transit system. In the modern production line, robots can greatly
improve processing efficiency. Taking the production of a high-speed train gearbox as an
Robot Systems for Rail Transit Applications. https://doi.org/10.1016/B978-0-12-822968-2.00001-2
Copyright © 2020 Elsevier Inc. All rights reserved.
1

example, in the process flow of a transmission gear, robot arms can result in the efficient
transmission of gears between different machine tools; when welding the gearbox, the
welding robots can improve machining efficiency; when assembling the gearbox, assembly
robots can work with high-precision; when the train is assembled, AGV enables the gearbox
to be transported quickly between workshops. Applying robots in manufacturing not only
saves processing time and labor costs, it improves manufacturing quality.
1.1.1.1 Assembly robots
As for assembly robots, the primary mission is to achieve high-precision positioning of the
workpiece. According to previous research, assembly costs account for 50% of total
manufacturing costs [5]. Assembly robot systems can also be divided into rigid assembly
and flexible assembly robots. Rigid assembly robots are customized processing systems for
specific workpieces in the traditional industrial environment.
Rigid assembly robots have poor generalization. If the production line is replaced with
processed parts, the equipment needs to be customized. Replacement of equipment will
cause a great economic burden. Compared with rigid assembly robots, flexible assembly
robots can design customized processing programs according to the workpiece. Flexible
assembly robots are programmable, which can result in different assembly schemes for
different workpieces. Flexible assembly robots are significant for a flexible assembly system.
In current industrial development, flexible assembly robots are the focus of development [6].
In the following discussion, assembly robots refer to flexible assembly robots.
Figure 1.1
Robots in the manufacturing, dispatch, and maintenance of rail transit.
2 Chapter 1

An assembly robot consists of four components: machinery components, sensors,
controllers, and actuators. To bring about a complex workpiece track in the real assembly
environment, assembly robots usually have more than four degrees of freedom (DOFs).
Mainstream assembly robots can be divided into two types: selective compliant assembly
robot arms (SCARAs) and six-DOF robots.
SCARAs have four DOFs, which are commonly used in electronic assembly, screw
assembly, and so on [7]. SCARAs are specially designed for assembly applications by
Yamanashi University. SCARAs contain two parallel joints, which can assemble a
workpiece in a specified plane. Compared with six-DOF robots, advantages of SCARAs
are a higher assembly speed and precision; disadvantages are limited workspace.
Commonly used control strategies for SCARAs contain adaptive control, force control,
robust control, and so forth [8]. In state-of-the-art research on robot control, intelligent
algorithms are employed to improve control performance [9]. Dulger et al. applied a
neural network to control the SCARA [10]. The neural network was optimized by particle
swarm optimization to improve performance. Son et al. adopted an optimized inverse
neural network for feedback control [11]. To deal with disturbances in running, the
parameters of the inverse neural network are updated by a back propagation algorithm.
Luan et al. used the radial basis function (RBF) neural network to achieve dynamic
control of the SCARA [12].
Six-DOF robots can locate the workpiece at almost any point. Thus, six-DOF robots can
handle the assembly task of complex three-dimensional (3D) workpieces. The dynamics
of six-DOF robots are basic for operating the robots. Zhang et al. considered the friction
of the robots and used a hybrid optimization method to model the dynamics of the
six-DOF robot [13]. After optimization, dynamic accuracy increased significantly. Yang
et al. proposed a simulator for the dynamics of the six-DOF robot [14]. Robots with a
large degree of freedom have large feasibility. However, too much freedom is
uneconomical. To handle the trade-off between economy and feasibility, the DOF can be
optimized for specific tasks. Yang et al. proposed an optimization method to minimize
the DOF [15]. This optimization method can reduce the DOF and improve the use of the
DOF.
Assembly robots should cooperate with the ancillary equipment. The fixtures are vital
equipment to ensure cooperation in performance. The fixtures can fix the relative position
between the workpiece and the robot under load. If the precision of the location of the
fixtures is low, no matter how accurate the positioning precision of the robot is, it cannot
achieve high-precision assembly. Currently, the flexible fixture is a future development
[16]. Lowth et al. proposed a unique fixture that can adjust the radial and angular
adaptively [17]. Although auxiliary devices are applied for assembly robots, the results of
assembly robots may still be unsuccessful. Avoiding unsuccessful assembly is particularly
Introduction 3

important in electric connector assembly, because the electric connector is not a rigid
component. To detect the unsuccessful assembly of the electric connector assembly, Di
et al. proposed a hybrid detection system with a force sensor and camera [18].
The fault diagnosis and prognosis system of assembly robots guarantees assembly
accuracy. There are many studies about fault diagnosis and prognosis systems. Huang
et al. designed a classifier for the wiring harness robot [19]. That study modeled the
manufacturing process was and calculated the fault with a fuzzy model. Baydar et al.
introduced a diagnosis model with error prediction [20]. The proposed model integrated
the Monte Carlo simulation, genetic algorithm, and so forth. The functions of the fault
diagnosis and prognosis system for assembly robots should contain the main aspects as
given in Choo et al. [21]:
(a) The health states of the assembly robots are monitored in real time. The monitored
data are logged into the dataset. The health features are extracted from the health
states of the assembly robots. The faults and remaining useful life can be calculated
according to the features.
(b) According to the fault diagnosis and prognosis results, the assembly tasks are reas-
signed to make sure the failed assembly robots are replaced by the fully functioning
robots. The maintenance plans can be made to repair the failed robots.
1.1.1.2 Collaborative robots
When applying the resulting classical industrial robot systems, interaction between robots
and humans is limited. There are three reasons for this phenomenon:
(a) Traditional industrial robots do not consider moving humans. If a collision occurs along
a certain trajectory, it may cause great damage to humans. Therefore, the working area
of the industrial robot is mostly separated from the working area of the human.
(b) The weight and volume of traditional industrial robots are large, and it is difficult for
humans to operate robots.
(c) Reprogramming of robots is difficult and requires special programming tools for
tuning.
However, humanerobot collaboration can combine human creativity with the efficiency of
robots and can amplify the flexibility of robots and further improve work efficiency. Under
this demand, collaborative robots are born. Compared with traditional robots, collaborative
robots have three main advantages: safety, ease of operation, and ease of teaching. Some
robot manufacturers have launched collaborative robotics products. The robot company
Universal Robot launched the UR3 collaborative robot [22]. This robot is the first truly
collaborative robot. The UR3 collaborative robot is based on a six-DOF robotic arm and is
flexible enough to achieve complex motion trajectories. In terms of safety, the UR3
4 Chapter 1

collaborative robot has a collision monitoring system that protects human safety by
monitoring the joint position, speed, and power of the robot.
KUKA Robotics has launched a collaborative robot, the LBR iiwa [22]. The robot’s seven-
DOF design provides greater flexibility than traditional six-DOF robots, enabling more
complex trajectories to cope with complex environments that work with humans. The
shape of the robot is designed to be ergonomic and easy for humans to operate. The outer
casing is made of aluminum alloy, which can reduce weight and improve operability. The
robot is equipped with torque sensors at each joint to monitor collisions in real time. The
teaching method of the robot is dragging, which reduces the technical threshold of the
robot operator.
ABB launched the robot YuMi [22]. The robot is highly safe and can achieve
humanerobot interaction in a small space. To improve the performance of collaborative
robots, AAB acquired Gomtec Robotics, which launched the collaborative robot Roberta
[22]. The Roberta can handle higher load applications compared with YuMi.
Franka Emika launched the Franka Collaboration [23]. Like the LBR iiwa, the robot has
seven DOFs. It is also equipped with a torque sensor on each joint to enable collision
monitoring.
Rethink Robotics launched the two-arm collaborative robot Baxter and the one-arm
collaborative robot Sawyer [24]. The two robots are exquisitely designed with high
positioning accuracy and can be assembled with high precision.
Safety in humanerobot interaction is essential for collaborative robots. Threats to the
safety of collaborative robots can be divided into two aspects [25]:
(a) The first kind of threat is from robots. During operation, the robot may collide with
workers and cause injuries. To ensure the safety of employees, the robot needs to
detect the location of workers in real time and determine whether the location of
workers is in a safe position. If workers intrude on the safe area, the robot should
immediately stop to avoid a collision. In case a collision between a person and a robot
occurs, the robot needs to detect the collision in time and change torque to minimize
damage to workers. Mohammed et al. proposed a collision avoidance system for
collaborative robots [26]. This system used depth vision sensors to detect the position
of the worker. Considering the virtual model of the collaborative robot, a collision
could be detected. The collaborative robot could take measures to avoid collisions. In
addition to collision detection, it is necessary to ensure the integrity of the collabora-
tive robot control system during operation. Failure of any part of the sensors, control-
lers, or actuators will lead to the failure of humanecomputer interaction, thus
threatening the safety of workers. In addition, interaction with robots may cause
mental stress to workers [27], which will increase the risk for a collision.
Introduction 5

(b) The second kind of threat is from the industrial process. In the process of humane
robot interaction, workers need close contact with the manufacturing process. The
temperature of the manufactured workpieces can cause damage to workers. Fallen
workpieces can also endanger workers. Therefore, it is necessary to fully consider the
impact of machining parts on workers in the design process of collaborative robots. In
addition, the unreasonable ergonomic design of the collaborative robot during mainte-
nance will have an impact on safety.
1.1.1.3 Automatic guided vehicles
According to the level of automation, manufacturing systems can be divided into three
levels [28]. Manufacturing systems in the first level are manual. Those in the second level
have small-scale automated manufacturing in which transportation is carried out manually.
Manufacturing systems in the third level have large-scale automated manufacturing and
use automated transportation. Manufacturing systems in the third level are also called
flexible manufacturing systems (FMSs). According to the Material Handling Industry of
America, only 20% of the time is spent on processing and manufacturing; the remaining
80% is used for storage, handling, waiting for processing, and transportation [29]. As
factory automation increases, transportation efficiency between workstations needs to
improve. In the FMS, the AGV can improve the use of space in the factory and the
efficiency of transportation in the material handling system. Therefore, transportation costs
can be reduced.
The rail transit manufacturing system is a typical FMS. Transportation is an important part
of the rail transit manufacturing system. In the rail transit processing environment, not
only the transfer of workpieces between processes within the plant but also the free flow
of workpieces between plants is required. In the traditional rail transit manufacturing
environment, transportation inside the factory is carried out by gantry cranes and the
transportation between factories is carried out by trucks. These modes of transportation
have some disadvantages. There is a safety hazard when using gantry cranes to lift. If the
lifting workpiece falls, it may cause serious safety accidents. The use of trucks carries a
higher cost and is less efficient to transport. Improving the safety and economy of products
in the transportation process is important for the development of the rail transit industry. In
the current manufacturing environment, the AGV is an effective mode of transportation.
The AGV is safer compared with gantry cranes and is more automated and more efficient
for transportation than trucks.
The typical structure of an AGV consists of sensors, chassis, a control unit, and so on
[30,31]. During work, the sensor can determine the position of the AGV and transmit the
current position to the control decision system, and the control decision system plans an
6 Chapter 1

optimal path. The chassis is driven by the controller’s control command to transport the
workpiece to the designated position.
To improve transportation performance, the model predictive control of the AGVs should
be achieved. The remaining power and the working state of the AGV should be predicted
to calculate the remaining life of the AGV, and the dispatch plan of the AGV can be
optimized with consideration of these factors to improve the operational safety of the
AGV. Popular data-driven forecasting methods contain statistical methods, intelligent
methods, and hybrid methods. The statistical methods can discover the statistical rule of
the data and generate an explicit equation for prediction. Commonly used statistical
methods contain autoregressive moving average, Winner process, Gaussian process, and so
on. Intelligent methods can generate better forecasting performance than statistical
methods with the help of the strongly fitting capacity of the neural network. The Elman
neural network, multilayer perceptron, and extreme learning machine (ELM) are the three
most popular intelligent methods. However, the training process of these neural networks
depends on the initial values in some way. If the initial value is unsuitable, training of the
neural network may stop at the locally optimal solution. To improve the performance of
intelligent prediction methods, the initial values can be optimized by optimization
algorithms [32]. Hybrid prediction methods combine data processing algorithms with
statistical or intelligent methods. The decomposition algorithms are proved to be effective
[33]. The decomposition algorithms can divide the raw series into several more stationary
subseries. Each group of subseries has a simpler fluctuation mode than the raw series, so it
is more predictable.
After optimization, the control command can be assigned to the AGVs in two different
ways: static control and dynamic control [34]:
(a) Static control. The control commands are assigned before the task. Once the AGVs
receive the control command, the transportation path will not be changed until the
AGVs receive another control command. This control scheme is simple and easy to
operate. However, flexibility is weak.
(b) Dynamic control. This control method can adjust the control commands according to
the real-time state of the AGVs, so the task scheduling strategy is complex.
The multi-AGV system is being studied worldwide. Compared with the single AGV, the
advantages of the multi-AGV system are:
(a) The multi-AGV system can cover a large area. The single AGV can achieve trans-
portation only between points. In the modern manufacturing environment, the
transportation task is far more complex than the point-to-point transportation. The
multi-AGV system can build a transportation network and improve transportation
efficiency.
Introduction 7

(b) The multi-AGV system can execute the transportation task in parallel. A multi-AGV
system can perform tasks simultaneously for a complex task. Thus, the multi-AGV
system can greatly improve the efficiency of transportation.
The dispatch and routing of the multi-AGV are important. The conflict-free function of the
multiple AGVs is the bottleneck for the multi-AGV system. Draganjac et al. proposed a
control algorithm for the multi-AGV system [35]. The proposed algorithm can detect the
conflict between the AGVs and guarantee the safe operation of the multi-AGV by the
priority mechanism. Miyamoto et al. proposed a conflict-free routing algorithm for the
multi-AGV system [36]. Because of the limited memory space of each AGV, a heuristic
algorithm was adopted for routing. Małopolski et al. considered the transportation system
in the factory as a combination of squares [37]. Based on the square topology, a novel
conflict-free routing algorithm for the multi-AGV system was proposed.
The dispatch plan of the multi-AGV should be calculated to optimize the task waiting
time, collision, load use, and so forth [38]. The optimization methods can be divided into
single- and multiple-objective optimization. Single-objective optimization can optimize
only one objective function. If the objective function consists of several objective
subfunctions, these subfunctions should be combined as the weighted sum [39]. However,
it is difficult to design these weights. Therefore, the generated optimization results might
not be the global optimal solution. The multiple-objective optimization can balance the
trade-off between different subfunctions and generate a Pareto front [40]. The Pareto front
contains many solutions, each of which has both advantages and disadvantages. The final
optimization results should be selected according to the expert. In this manner, the
intelligent decision-making ability of humans can be used to improve the dispatch
performance of the AGV.
The environment of the factory is dynamic. The AGVs should be flexible enough to cope
with the dynamic manufacturing system. There are many studies on the dynamic
transportation system. Brito et al. proposed a dynamic obstacle avoidance algorithm for
the dynamic unstructured environment [41]. In this algorithm, model predictive control is
applied to improve control performance. This algorithm was verified in the environment
with walking humans. Li et al. proposed an integrated algorithm for obstacle avoidance
[42]. This algorithm can generate a path for the AGV by a model-predictive algorithm.
The AGV can track the generated path reliably.
1.1.1.4 Manufacturing robots
Manufacturing robots contain many types including welding robots, drilling robots,
grinding robots, milling robots, and so on. Manufacturing robots can produce better
machining performance than a classical computerized numerical control (CNC) machine.
8 Chapter 1

For example, it is proven that a workpiece polished by a manufacturing robot has a better
surface quality than any CNC machine [43]. The better performance of the manufacturing
robots is because the robots have more flexibility to make sure the tool is in the right
position.
Welding robots are one of the most widely used manufacturing robots. Difficulties of
welding robots are [44]: (1) it is hard to observe the welding seam in a complex
manufacturing task, (b) it is hard to obtain the absolute and relative locations of the
workpieces, and (c) the trajectory is hard to track. Current commonly used location
methods for the welding seam are based on optical sensors such as a depth camera and
laser sensor [45,46]. Jia et al. proposed a welding seam location method and trajectory
tracking algorithm [44]. The proposed location method was achieved using a laser scanner,
which could obtain the location and direction of the pipe. The cubic spline was applied to
fit the obtained welding seam. The velocity control was used to track the welding
trajectory. Liu et al. proposed a trajectory planning algorithm for welding robots to cope
with a single Y-groove welding task [47]. This study provided two different velocity
planning algorithms.
A key concern with drilling robots is positioning accuracy. An inaccurate hole can reduce
the mechanical performance of the equipment. The positioning compensation method can
be divided into two types: model-based and model-free [48]. Model-based methods can
guide the robot to move according to the measured positioning error. The essentials of
model-based methods are the dynamics and kinematics of the drilling robots. Model-based
methods are time-consuming. Model-free methods can solve this drawback. Model-free
methods build a model to describe the relationship between the positioning error and the
robot’s joints’ parameters, which do not consider the dynamics and kinematics of the
robots. Commonly used model-free methods contain the interpolation method [49],
cokriging method [50], and so on. Neural networks are applied to compensate intelligently
for positioning . With the help of the strongly fitting capacity of neural networks, these
intelligent positioning compensation methods obtain good performance. Yuan et al. used
ELM to predict positioning error and guide the robot to compensate for it [51]. Chen et al.
used the RBF neural network to estimate positioning error. The bandwidth of the adopted
RBF neural was fine-tuned. Positioning accuracy can be improved by more than 80% with
these intelligent positioning compensation methods [52].
The grinding manufacturing is highly precise, so the positioning accuracy of the grinding
robots requires attention. In real applications, the grinding robots may deviate from the
preset track because of the disturbance and wear to the tools. Thus, grinding robots should
have the ability to self-adjust, to ensure manufacturing quality. Control of grinding robots
is more difficult than for CNCs because grinding robots have more DOFs. Huang et al.
proposed an intelligent gear grinding system [53] in which the robot arm can detect the
Introduction 9

actual trajectory by a vision sensor and adjust the trajectory adaptively. Two cameras are
adopted for tool center point calibration. When coping with the complex grinding task,
multiple grinding robots are necessary, because multiple grinding robots can cope with the
complex manufacturing task more easily than a single grinding robot. Han et al. [54]
proposed a multiple grinding robot system and introduced the trajectory planning method
for the multiple-robot system. Experimental studies indicated that the multiple grinding
robot system can generate a steadier manufacturing trajectory.
The robot only needs five DOFs for the milling task; the reserved one is the DOF of the
spindle of the milling cutter [55]. In the real application, six-DOF robots are applied for
milling to improve flexibility. The stiffness of milling robots is a challenge in
manufacturing. Milling robots usually have low stiffness. Robots may deform or vibrate
when milling workpieces. There have been many studies on improving milling
performance. Peng et al. optimized the stiffness of the robots in the feed direction with a
seven-DOF robot [56].
Commonly used manufacturing robots are developed from six-DOF robots. Trajectory
planning is a common technology for manufacturing robots. The trajectory planning
algorithm can generate the optimal trajectory and the robot control algorithm can drive the
robot to follow a predetermined trajectory [57]. The optimization targets of the trajectory
are the position, velocity, accelerated velocity of each joint. The trajectory planning
algorithm is the basis of the industrial robot control system. The optimization functions of
the trajectory contain many aspects, including reduce execution time, consumed energy,
and impact.
1.1.1.5 Loadingeunloading robots
In the existing transportation manufacturing environment, processing each workpiece
requires different procedures. The connection between processes requires the workpiece to
be moved from one machine to another [58]. This repetitive work can be replaced by
loadingeunloading robots. The application of loadingeunloading robots has many
benefits. On the one hand, loading and unloading of robots is repetitive and consistent,
which can avoid a decrease in work efficiency caused by worker fatigue. On the other
hand, using the robots can parameterize the loading and unloading operation and the fault
source can be quickly located when troubleshooting problems, which is convenient for
improving the processing quality.
Loadingeunloading robots have a wide range of applications. Liu et al. adopted a
loadingeunloading robot for the CNC and designed a control system based on a
programmable logic controller [59]. Zhang et al. developed a robot system for
loadingeunloading [60]. The designed robot was able to separate good and bad
workpieces. Fan et al. used the loadingeunloading robot for electric equipment detection
10 Chapter 1

[61]. With the help of the loadingeunloading robot, automatic and high-precision
detection is achieved.
The development of loadingeunloading robots shows a trend in intelligence. Computer
vision and neural network technology have been widely applied [62]. The adoption of
computer vision can increase the flexibility of loadingeunloading robots.
Loadingeunloading robots can observe the posture of the workpiece and adjust the robots
to grasp the workpiece from the appropriate position. The neural network is widely used in
controller design with its strong generalization ability. Gu et al. proposed a visual servo
loadingeunloading robot [63]. The camera on the manipulator can detect the location of
the workpiece and help the robot capture the workpiece precisely. The design contains two
steps: (1) extract the features of the observed image and target image and calculate the
error between these features; and (2) input the error into the neural network and output the
control command for the robot. In the proposed robot control structure, the fuzzy neural
network was adopted as the controller.
1.1.2 Rail transit robots in dispatch
The running stage of the rail transit system takes up the longest length and generates the
highest cost in the whole life cycle of the rail transit system. With the help of robot
technology, the rail transit system can achieve automatic operation, thus improving
operational safety and reducing transportation costs. Autonomous driving is the most
important form of automation in rail transit. Autonomous driving is the application of
robot control technology to the railway train. The automatic train control of the railway
train contains automatic train supervision (ATS), automatic train protection (ATP), and
automatic train operation (ATO) [64]. The ATS can supervise the running states of the
railway train. The ATP system can monitor the train’s running position and obtain its
speed limit to ensure its running interval and safety. The ATO can accept output
information of the ATS and ATP and generate control instructions [65].
The international standard International Electrotechnical Commission 62,290 defines four
grades of automation (GOA) [66]:
(a) GOA1. The train is able to monitor the train’s operating status continuously. Operation
needs to be carried out by drivers.
(b) GOA2. This level is self-driving with driver duty. The train is able to drive automati-
cally through the signal system, but the driver is required to close the door and issue
the command.
(c) GOA3. There is no need to equip drivers on this level of the train; the train can auto-
matically complete the whole process of operation, including outbound, pit stop,
Introduction 11

switch door, and so on. However, there is still a need for onboard personnel to deal
with emergencies on this level of train.
(d) GOA4. No operator is required on this class of trains, and the train’s control system
can automatically respond to unexpected situations.
ART is an important example of the automation of rail transit operation; it is developed by
CRRC Zhuzhou Institute Co., Ltd. The core technology of this train is virtual orbit
tracking control technology. That is to say, the ART train does not need a true track, but
rather, the marks on the road serve as tracks. Because the intelligent rail train does not
have a physical track, existing positioning devices used in the subway train and high-speed
train cannot be used. The ART adopts the global satellite positioning system as the main
positioning method [4]. During the operation, inertial sensors and angle sensors installed
in intelligent trains are used to monitor the movement posture and position of the train,
thereby enabling a comprehensive perception of the state of the ART train. The monitored
train status information is fed back to the controller of the intelligent rail train to control
the intelligent rail train to run along the virtual track. Using this control technology, all
wheels of the ART train can be driven along a virtual track [4]. This position accuracy can
ensure the passing performance of the ART rail train in urban traffic. To improve
operational efficiency, the ART train is also equipped with automatic driving technology
[4]. The ART train can intelligently sense the environment and control the automatic
operation of the train.
1.1.3 Rail transit robots in maintenance
Robots have become the focus of modern rail transit equipment manufacturing and operation.
With the increase in railway operating mileage and operational density, the workload of
railway maintenance support is also increasing, and higher requirements are put forward for
the maintenance and repair of railway infrastructure equipment. Currently, the maintenance of
railway infrastructure equipment is mainly carried out by workers [67]. The main problems
Figure 1.2
Role of robots in rail transit maintenance.
12 Chapter 1

are that (1) diversified test operations have higher requirements for operators, (2) it is
inconvenient to have operators carry measurement equipment, and (3) manual inspections
maintain labor intensity and quality is uncontrollable. As shown in Fig. 1.2, robots can have
an important role in the maintenance stage of rail transit in this case.
1.1.3.1 Inspection robots
The objectives of railway maintenance include tracks, instruments, and equipment next to
railways, traction substations, pantographs, temperature of instruments, bogies, and foreign
matter intrusion. According to the work scenario, the inspection area of rail transit
inspection robots is mainly divided into traction substations and railway.
1.1.3.1.1 Inspection in traction substation
The traction substation converts electric energy from the regional power system into
electric energy suitable for electric traction according to the different requirements of
electric traction on current and voltage. The resulting electricity is then sent to catenary
lines set up along the railway line to power electric locomotives, or to the urban rail transit
system to power electric subway vehicles. The inspection work is the top priority for the
daily maintenance of the traction substation. However, because of insufficient staffing and
other reasons, the inspection cannot meet the needs. In addition, with the increase in
substations, operation, and maintenance, personnel are struggling to comply. It is also
difficult to keep satisfactory working conditions and a strong sense of responsibility.
Currently, the technology of intelligent robots is increasingly developing and the use of
robots to replace manual inspections of equipment partially or completely has become the
trend of future substations [68]. Substation intelligent inspection robots are equipped with
advanced equipment such as infrared thermometers, high-definition cameras, and audio
signal receivers. According to the preset inspection tasks, the entire station equipment and
partitions are inspected in time. All task modules of meter reading, temperature
measurement, sound collection, and functional inspection are processed together to
identify and alert regarding discovered equipment abnormalities, defects, and hidden
dangers in a timely manner. In addition, robots intuitively form charts and report weekly
on operational data indicators of devices stored in the inspection so that maintenance
personnel can perform equipment maintenance and fault analysis better.
By using the inspection robots in the substation, the equipment in the substation can be
inspected more reliably and the data analysis returned by the system can give the
maintenance personnel a better understanding of the working state of the equipment.
Recording and analyzing historical data can also significantly improve the predictability of
equipment in the substation, providing more effective data support for the condition of
maintenance and a status assessment of the equipment, which will directly reduce the
possibility of equipment accidents [69]. Using inspection robots to inspect substation
Introduction 13

equipment can not only keep field maintenance personnel away from the field equipment,
which improves the safety and reliability of the inspection work, it provides a new method
for spot inspection for the popularization of an unattended substation and truly improves
the safety and reliability of substation equipment.
1.1.3.1.2 Inspection along the railway
With the rapid development of rail transit and the continuous improvement in passenger
high-speed processes, railway transport has been gradually heading toward function
integration, information sharing, command concentration, and highly automated transition
and transformation. The urgent demand for railway transportation safety poses new
challenges to the railway traffic safety assurance system. To adapt to the new needs of
railway development, it has become the focus of the construction of the railway traffic
safety assurance system to build a fully covered and highly reliable railway intelligent
monitoring system based on high information sharing [70]. The higher speed brings
convenience to people, but it also increases the difficulty of preventing railway disasters.
Although a high-speed train is equipped with a more advanced safety system, the large
braking distance caused by high-speed driving reduces the flexibility of the train to
respond to emergencies. In the event of an accident, a high-speed train often has more
serious consequences than an ordinary train. Therefore, various factors such as the
intrusion of foreign matter, natural disasters, and line failures may cause major railway
accidents. Casualties and property loss caused by the intrusion are enormous and
frequently occur. Therefore, research and development into railway foreign matter
intrusion detection are of great significance to ensure the safe operation of trains [71].
The intrusion of foreign matter has the characteristics of irregularity, suddenness, and
unpredictability. The fixed transportation mode of the operation route makes emergency
measures that can be adopted limited when the train detects an intrusion on the road. If
real-time detection of invasive foreign matter can be achieved and necessary measures are
taken in time, property loss can be reduced as much as possible. The traditional foreign
matter intrusion approach is to place sensors at key positions along the railway to monitor
intrusion conditions in real-time. This method has many drawbacks, such as an
unreasonable arrangement of the sensor and fixed monitoring position. With the rapid
development of robotics, robots have entered the field of railway foreign matter invasion
detection. The railway is inspected by an infrared sensor, a laser sensor, a smart camera,
and an ultrasonic device.
1.1.3.2 Channel robots
High-speed trains may experience poor performance or even failure of key components
during long-term high-speed service. If a safety accident occurs, it will cause severe
economic loss and even casualties. Therefore, how to monitor and diagnose its operating
14 Chapter 1

Another Random Scribd Document
with Unrelated Content

"I shall, I hope, be always as kind-minded to you as I am now."
"And never any more?"
"I cannot be any more."
"Is there--is there no hope?" She clasped her hands and looked
up at him in wild appeal.
He shook his head. "I loved you once, but I cannot love you
again."
"You say you forgive me. If you forgive me, why cannot you love
me? for I love you now as I never loved any one before."
"Too late! Too late!"
"Is it because my good looks are gone? Why, O, why cannot you
love me again, unless it is because my good looks are gone?"
"No; your good looks have no weight in the matter. I could not
forgive you if I loved you in the old way."
"Then," she cried, rising and stretching forth her arms wildly
towards him, "do not forgive me; revile me, abuse me, yes, beat me,
but tell me you love me as you did long ago; for I love you now
above anything and all things on earth. Yes, ten thousand times
better than I love my child! I never knew you until now. I was too
giddy and vain and shallow to understand you. I have behaved to
you worse than a murderess. But, Frank, I would die for you now!"
She flung herself on her knees on the floor, and raised her clasped
hands above her streaming face to him. "On my knees I ask you in
the name of merciful Heaven to give me back your love, as I had it
once! Give it to me for a little while, and then I shall be content to
die. You are noble enough to forgive me and to take me back into
your house. Take me back into your heart too. Raise me up and take
me in your arms once, and then I will kill myself, if you wish it; I

shall then die content. Refill my empty veins with words of love and
I will trouble you no more. I have been walking blindfold in the
desert all my life, and now that the bandages are taken off my eyes
and I can see the promised land, am I to find I can never enter it? I
am only a weak, wicked woman. You have extended to me
forgiveness that makes you a god. Have for me, a weak woman, the
pity of a god."
"I am no longer a man," he said, leaning against the wall. "I am
smoke, an abstraction, a thing, an idea, a code. You are my wife and
I will not cast a stone at you. You are my wife, and you are entitled
to the shelter of my roof and the protection of my name. I make you
free of both. But when you ask for love such as once was yours, I
fail to catch the meaning of your words. You are speaking a
language the import of which is lost to me. It is not that I will not,
but that I cannot, give you what you ask. There would have been no
meaning in the love I offered you years ago if I could offer you love
now. Get up. It was with a view to avoiding a scene I spoke."
"I will not get up until you tell me there is hope--that some day
you may relent."
"There is no question of relenting. When you left me you
destroyed in me the faculty of loving you. Now get up. We have had
enough of this. We must have no more. I have been betrayed into
saying things I determined not even to refer to. Get up, and, mind,
no more of this." With strong, firm arms he raised her from her
knees.
She stood for a moment, leaning one hand on the table to steady
herself. Then in a low quavering whisper, she said, "Is there any, any
hope?"
"There is none."
She raised herself, and moved with uncertain feet to the door. "It
would have been better I went to the river last night."

CHAPTER XXXVI.
TEA AT CRAWFORD'S HOUSE.
When Philip Ray left Boland's Ait he crossed over to the tow-path,
and not to Crawford's Quay. It was still too early to call at Layard's.
There was nothing else for it but to kill time walking about. Under
ordinary circumstances when greatly excited he went for a very long
walk. If nothing else but the startling and confounding affairs at
Boland's Ait had to be considered, he would have dashed off at the
top of his speed and kept on straight until he had calmed himself or
worn himself out. But there was Crawford's House to be thought of.
That must not be left far behind. Even now when he intended
circling it he could not bear to think he was turning his face away
from it, although he knew it was necessary to make a radius before
he could begin his circle.
His mind was in a whirl, and he could see nothing clearly. The
astounding return of his sister from the grave, and the still more
astounding pardon extended to her by her husband, threw all his
ideas into phantasmagoric confusion. Images leaped and bounded
through his brain, and would not wait to be examined. Of only one
thing was he certain: that Frank was the noblest man he had ever
met. Although he repeated over and over to himself Bramwell's
words about Kate, although over and over again he called up the
vision of Kate in that room on the islet, he could not convince his
reason that forgiveness had been extended to her. In his memory he
saw the figures and heard the voices, and understood the words
spoken, but a dozen times he asked himself, could it be true? or had
his imagination played him false?

The affairs at the Ait dwarfed his own concerns, and made them
seem tame and commonplace. That a young man should fall
desperately in love with a beautiful girl like Hetty was the most
natural thing in the world; but that a hermit, a young man of
scrupulous honour like Frank, should take back an errant wife,
whose former beauty had now turned almost to repulsiveness,
transcended belief. It was true, but it was incredible.
As time went on, and the walking allayed the tumult in his mind,
his thoughts came to his own position in the circumstances. He had
not told Layard or Hetty any of Frank's history beyond the fact that it
was a painful one, and a subject to be avoided. He had not told
them that he was Bramwell's brother-in-law. He had never said a
word about Bramwell's wife.
Now all would have to be explained. Of course, he had intended
telling when he spoke to Layard about Hetty; things had changed
beyond anticipation, beyond belief, since last night. Had he known
what was going to happen on the Ait last night, what had absolutely
happened when Hetty and he landed there, he would not have said
a word of love to the girl. He would have told her the facts about
Kate before asking Hetty to marry Kate's brother, before asking Hetty
to become the sister of this miserable woman.
He knew he was in no way responsible for his sister's sins, but
some people considered a whole family tainted by such an act in one
of its members. Some people believed conduct of this kind was a
matter of heredity, and ran in the blood. Some people would ask, If
the sister did this, what could you expect from the brother?
Would the painful tale he had to tell Layard influence Hetty's
brother against his suit? There were thousands of people who would
consider that he himself was smirched by his sister's fault. Was
Layard one of these?

The best thing for him to do was to relate the story at once; the
most honourable and straightforward way for him to proceed would
be to speak to Layard before he again saw Hetty. If Layard raised an
objection, and that objection was insuperable, the most honourable
course for him to pursue would be to give up all pretensions to
Hetty.
Yes, but could he? And would he be justified in renouncing her
now that he knew she loved him? It would be all very well if he had
not made love to her and gone so far as to ask her to marry him. If
only his happiness were concerned the path of duty would be plain
enough. But Hetty and he were now partners in love, and had he the
power or the right to dissolve the partnership without consulting
her? Clearly not. However he looked at the situation doubts and
difficulties arose before his mind. There was only one matter clear--
he ought to speak to Layard at once.
It was now half-past seven. Layard left the gasworks at eight.
Why should he not intercept him on his way home and put him in
possession of all the facts? Upon what Layard said, the course to be
adopted could be based.
He got to the gas-house, and was walking up and down
impatiently when Alfred Layard came out of the gateway and saw
him.
"Anything the matter?" asked Layard apprehensively when Ray
came up to him.
"At your place? O, no! I wanted a few minutes' talk with you, so I
came to meet you."
"All right," said Layard, with a smile. He thought he could guess
what the talk would prove to be about. He was the incarnation of
unselfishness, and it never occurred to him for a moment to consider
how awkward it would be for him if Hetty married and left him.

"I want first of all to tell you a very painful piece of family
history," said Ray, anxious to get the worst over as soon as possible.
"But why should you, Ray? I am the least curious man alive."
"You will know why I wish to tell you before I have finished."
Then, without further preface, he narrated the history of Kate, her
marriage, her flight, her supposed death, her appearance last night
at the Ait, and her husband's forgiveness.
Layard was greatly interested and excited by the story. When it
was finished, he said:
"There is enough Christianity in that man Bramwell to make a
bishop."
"To make the whole bench of bishops," cried Ray enthusiastically.
"I always knew he was a hero, but I was not prepared to find the
spirit of a martyr as well. And yet I ought to have been prepared for
anything noble and disinterested in him. He does what he believes to
be right without any view to reward here or hereafter. He has had
his wild days when he plunged, under his great trouble, into the
excitement of gambling, but even in that he was unselfish; he
injured no one but himself. Once he pulled up, he stopped for good
and all. And now I come to the reason for taking you into confidence
and telling you what you need never have known only for something
which concerns myself more deeply than all else which has
happened to me in my life."
Then in a few words he explained his position, his feelings
towards Hetty, and his belief that his feelings were reciprocated.
"You have three matters to weigh," he said, in conclusion; "first,
the family history I have told you; then my financial position, taking
into account the chance of my getting the tuitions; and, last,
whether you would object to me personally. In the short time I have

known you, I have taken to you more than to any other man I ever
met except Frank. I am speaking to you as much as a friend as
Hetty's brother. If I did not look on you as a friend, I should not care
greatly to take you into my confidence and defer to you. But the
notion of doing anything underhand or behind your back would
seem to me intolerable treason."
"I'll be as straightforward with you as you have been with me. I
have liked you from the first moment of our short acquaintance. The
way in which you have spoken to me this evening strengthens ten
thousand times my good opinion of you. The miserable family
history you have told me has no bearing whatever on you, and I see
nothing to stop you but the getting of those tuitions. Why, I married
on little more than your salary; and during my short married life I
never for one moment repented, nor did my poor girl. Contented
and willing hearts are the riches of marriage, not money."
Ray was too much moved to say more than "Thank you, Layard;"
but he stopped in his walk, and, with tears in his eyes, wrung the
hands of the other man.
"And now," said Layard, as they resumed their way, "let us get
home to tea."
That was his way of telling Ray that there was no need of further
words either in explanation or of thanks.
"I thought we were going to have a thunderstorm last night, and
to-night it looks like it too. I always feel a coming storm in the
muscles of my arms, and they are tingling this evening."
Layard opened the door with his latch-key. The two men went
into the front room, and in a few minutes Hetty appeared with the
tea-pot. She coloured deeply on seeing Ray with her brother. She
had not heard the footfalls of two people, and was not prepared to
find him there. He had never before come in with Alfred, and a
suspicion of what had occurred flashed through her mind.

She did not speak to Ray. She felt confused, and half-pretended,
even to herself, that she did not know he was present. Her brother
went to her and put his arm round her waist and kissed her cheek,
and then drew her over to the chimney-piece, where Ray stood,
feeling somewhat like a thief.
"You forgot to say good-evening to Ray," said the brother.
"Good-evening," said she, in a low voice, holding out her hand.
Ray took the long slender hand, feeling still more dishonest and
shamefaced and miserable.
When the fingers of the lovers touched, Layard caught the joined
hands in both his, and pressed them softly and silently together;
then, turning away, he stepped quickly to the window, and stood a
long time looking at the dead wall opposite through misty eyes.
"I don't think we shall have that storm," said Ray at length.
Layard turned round. Hetty was pouring out the tea, and Ray was
standing with his back to the chimney-piece.
"No," said Layard, "I fancy it is passing away. My arms feel
easier."
Hetty was smiling, but looking pale.
"Do you take sugar and milk, Mr. Ray?" said she.
"Dear me, Hetty," said her brother, "what a lot you have to learn
yet!"
She coloured violently, and shook her head at him.
"I wish you would sit down, Alfred. You are keeping all the light
out of the room; I can't see what I'm doing."

"No," said he, looking meaningly from her to Ray; "but, bad as
the light is, I can see what you have done."
At this Hetty and Ray laughed a suppressed laugh, and looked at
one another with joyous glances.

CHAPTER XXXVII.
CRAWFORD WRITES HOME.
The morning after Mrs. Crawford's relapse and Crawford's visit to
town about the three thousand pounds, the husband was sitting by
his wife's bedside. He was in a particularly cheerful and hopeful
humour, and insisted that she had already begun to mend, and
would in a week be better than she had been for months.
She shook her head with a sad smile, but said nothing. She did
not wish to sadden the being she loved above all other living
creatures by the thought of a final separation between them, a
separation which she felt was inevitable, and to which she could not
reconcile her mind. When alone she would cry out in despair to her
gentle heart, "To be so loved, and to be so loving, and to be
separated so soon!"
He went on affecting undiminished confidence in her recovery. "I
tell you, I am certain you must, you will, get well, and that much
sooner than even the doctor thinks." (The doctor had told him again
that day there was little hope of her rallying.) "What good would my
luck be if you were not by my side to share it? My Nelly comforted
and sustained me in my days of doubt and difficulty. Do you mean
that she is not to share my triumph? I will take very good care she
shall. And now I want to tell you what I insist upon doing. I will take
no denial, for I look on it as essential to your recovery."
"I will do anything you tell me," she said with meek devotion. "I
will do all I can to get well. For, William, I am the happiest and most

blessed woman in England, and I do not want to leave you, dear."
"That's my own brave wife," said he, winking his eyes quickly and
patting her arm. "I don't think you will raise much, if any, objection
to what I am about to do. I am going to write to Mrs. Farraday to
come back and stay with you. She promised she would come if you
needed her, and she will be a great source of comfort and
confidence to you."
"But her brother?"
"Oh, her brother can do without her for awhile. You will be all
right again in less than no time, and then, if she wishes, she can go
back to her brother. And now I am off to write to Rochester for her
by this very post, for a good thing cannot be done too soon, and I
am sure this is a good thing for you."
He left her, went to his own room, wrote the letter, and posted it
immediately himself. Then he came back to the house, and having
entered the dining-room on the ground-floor, began walking up and
down with brows lowered in deep meditation.
"I had better get it over me before Mrs. Farraday comes," he
thought as the result of his cogitations. "I can't stay here any longer.
I am not a sick-nurse to philander after an ailing woman, and dally
in an invalid's room. She was a fool to marry me. Did she think for a
moment I fell a victim to her ancient charms? If she did she ought to
be in a lunatic asylum. Of course I told her I wanted to marry her for
love, but is there in the history of the whole human race a single
case of a man saying to a woman, 'I want to marry you for your
money'? Not one.
"I can't stand this house any longer; it suffocates me. The doctor
says there is no hope. Why should I wait to see the end? The
approach of death and the presence of death are abhorrent to all
healthy people. I can do no good by staying, and I have to think of
myself. There are very few men living who would have been as good

to her as I have been. She cannot expect me to do more, and," with
one of his short laughs and a quick winking of his eyes, "my affairs
in South America urgently demand my presence. I'll get the business
over me at once. Brereton told me I could have the money early this
afternoon."
Here his mind became so intensely occupied that his legs ceased
to move, and he stood in the middle of the room lost in thought. He
was contemplating scenes in his imagination: not proceeding by
words. Presently words began to flow through his brain again, and
he resumed his pacing up and down.
"If there should be any hitch about that money I should be in a
nice mess." He shook his head gravely and repeated this
contingency to himself two or three times. "That would never do. It
would look weak and foolish. When I act I must act with firmness
and decision. No, I had better make sure of the cash first."
He put all the money he had in his pocket, left the house, and
took the first train to town. At Waterloo he jumped into a hansom
and drove straight to the office of Mrs. Crawford's solicitor. He found
Mr. Brereton in, and everything ready. The solicitor handed him an
open cheque for £3,270, saying gravely as he did so:
"And you are fully resolved to put this money in that South
American speculation?"
"My dear sir, there's a vast fortune in that fibre of mine; and now
that the machinery has been perfected, it is only stretching out one's
hands to gather in hundreds of thousands of pounds."
Brereton shook his head.
"The best place in which to put money is English Consols."
"What, less than three per cent.! For you can't buy even at par
now. Why, my dear sir, it's letting money rust."

"It's keeping money safe."
Crawford shrugged his shoulders and made a grimace of
dissatisfaction.
"Over-prudence, my dear Mr. Brereton. Who never ventured never
got."
"A bird in the hand is worth two in the bush; and of all the
uncertain things I know of there is only one worse than putting
money in South American speculations, and that is putting it in
Central American ones."
"Ah, but you have never been in South America!" said he
triumphantly, and his eyes winked quickly, and he laughed a short
unpleasant laugh, and thought to himself, "Nor have I either." Then
he continued aloud, "I am aware that it is most unwise of any one
who does not know the ground to dabble in South American
speculations, but, you see, I am well acquainted with the place, and
know the ropes."
"The last client I had who touched anything in South America
blew his brains out. But, of course, it is no affair of mine. I have only
to do what I am asked by Mrs. Crawford in her letter to me. The
cheque is an open one, as you requested. They will pay you across
the counter. I hope you will not think of keeping such a sum as that
in your house?"
"O, dear, no! I am going to remit it at once to my agent. When
you see me next, Mr. Brereton," laughing and winking his eyes, "you
will congratulate me upon my spirit and success."
"I hope so," said the lawyer drily, and in a tone and manner which
plainly said he believed nothing of the kind would occur.
Crawford said good-bye and went straight to the bank, where he
got thirty-two one-hundred pound notes and seventy in fives.

He had never had so much money in his possession before. He
had never had any sum approaching it. Once or twice after a good
racing week in the old times he had been master of five or six
hundred, but three thousand two hundred pounds! It was almost
incredible! And it was all in cash! It did not lie in the cold obstruction
of any bank. It was not represented by doubtful I.O.U.'s. It was not
represented by shadowy entries in a betting-book. It was not
invested in any shaky securities. It was not manifested by abstract
entries in a ledger. The money was concrete and tangible, and lying
safely in his breast-pocket under the stout cloth of his coat. He could
take it out and count it now if he liked. That minute he could start
for Monte Carlo or St. Petersburg, Australia or Norway.
As he walked along the streets he held his head high. He felt
independent of all men, independent of fortune, of Fate. He had
married for money, he had realised the prize, and it was now safe in
his pocket. These notes were as much legally his own as his hands
or his teeth. No one could take them from him except by force, and
he took pride in thinking that few men who passed him in the street
would be able to cope with him single-handed. He had as much
thought of risking his money in anything so far off and tame as
South American speculations as he had of buying a box of matches
and burning it note by note.
Of course, Brereton had been right in saying it would be a
dangerous thing to keep such a tempting sum in an ordinary house.
There might even be danger in walking about the streets with it in
his pocket. Some dishonest person might have seen him draw it out
of the bank and might be following him. He might be a match for
more than an average man, but he would be no match for two or
three. Garrotting had gone out of use, but it might be revived even
in midday in London by men who knew the prize he carried, and
were bold and prompt. If in a quiet street he were seized from
behind and throttled so that he could not cry out, and if a man in
front cut the pocket out of his coat, the thieves might be off before
passers-by knew what was going on or suspected anything being

wrong. He had a horror of revolvers, but plainly he ought to be
armed. He did not yet know where he should keep his hoard, but in
any case it would be well to possess the means of defending it.
Crawford had by this time got out of the City and was strolling
through Regent Street. He turned into a gunsmith's shop and bought
a short large-bore revolver and some cartridges. The man showed
him how to load the weapon. Crawford explained that he was about
to leave the country for Algiers, and wished to have all the chambers
charged, as he was going in a vessel with a crew of many
nationalities, and was taking out a lot of valuable jewellery.
Lying was a positive pleasure to him, even when it was not
necessary. "It keeps a man's hand in," he explained the habit to
himself.
It was now about two o'clock, and he began to feel the want of
luncheon. There was no place where better food could be got or
where the charges were more moderate than at the Counter Club.
He was only a short distance from it. What could be more
reasonable than that he should go and lunch there? Nothing. So he
turned into an off street on the left, and in a few minutes was
seated in a luxurious armchair in the dining-room, waiting for the
meal he had ordered of the obsequious waiter.
He was somewhat tired by his walk, and found rest in the well-
cushioned chair grateful and soothing.
Could anything be more comfortable and cheering than to sit at
ease in this well-appointed club, with a small fortune in notes under
one's coat? Here was no suggestion of illness or approaching death.
All the men present were in excellent health and spirits. They were
talking of cheerful subjects--horses, theatres, cards, the gossip and
scandal of the town. They spoke of nothing that was not a source of
enjoyment; and though all they said ran on assumption that they did
not contemplate the idea of any man denying himself pleasure or

being unable to obtain pleasure owing to the want of money, they
were not all rich men, but all spoke as if they were. It was so much
pleasanter to sit here, listening to this talk and taking part in it, than
to wander about that cold-mannered house in Singleton Terrace at
Richmond, or to sit by the sick-bed of a wife ten years older than
himself and whine out loving phrases and indulge in distasteful
private theatricals.
Then the obsequious and silent-footed waiter brought in his
cutlets, and whispered that his luncheon was ready. Everything was
very nice at Singleton Terrace, but somehow cutlets there and here
were two widely different matters. It was no doubt easy to explain
the reason of the difference. In one place the cook got twenty, in the
other a hundred, pounds a year. But though that explained the
difference, it made the cutlets at Singleton Terrace no better.
He had had enough of Richmond. Why should he go back there?
As he had always held, there was no advantage in being brutal, and
he would not undeceive his elderly wife. He would not tell her in
plain words that he had never cared in the least for her, that he had
married her merely for her money, and now that she was dying and
her income would, for him, die with her, and that he had got all the
money she had, that his whole mind was occupied with the image of
a beautiful young girl whom he was about to make love to and ask
to fly with him on her (his wife's) money. No. It would be uselessly
unkind to tell that middle-aged silly invalid any of these things. But
why should he go back to Richmond?
If he went back to say good-bye he would have to play a long
scene in private theatricals to which no salary was now attached,
since he had all the savings in his pocket. Besides, he would find it
hard, credulous as his elderly wife was, to make her believe there
could be any urgent necessity for his immediate departure to South
America. There would be a scene and tears--and he hated scenes
and tears--and then if the surprise or shock made her worse, who

could tell the consequences, the unpleasant consequences, which
might arise?
In the next room were pen, ink, and paper. Why should he not
write instead of going back? That was it! He'd write explaining, play
at the club to-night, and go on to Welford in the morning. That was
a better programme than crawling back to that silly old invalid and
acting sorrow at parting when his heart was overrunning with joy.
He went into the next room and wrote his first letter to his wife.
He used a sheet of unheaded paper, and did not date or domicile it.
My dearest Nellie,--Upon coming to town I found waiting for
me a telegram from Rio Janeiro to the effect that if I did not reach
that city at the very earliest moment possible--in fact, by a steamer
sailing from London to-day--my title to the estate on which the fibre
grows would lapse. Nothing but my personal presence could save it.
So, much against my will, I was obliged to drive in hot haste to the
boat without the satisfaction of bidding you good-bye. Indeed, I
have barely time to write this scrawl, and shall have to intrust it to a
waterman for post. Be quite sure all will go well with me, and that I
shall telegraph you the moment I land. I am so glad I wrote for Mrs.
Farraday before leaving home this morning. I know she will take
every care of my Nellie while I am away, and I am sure my Nellie will
take every care of herself, and be quite well long before the return
of her loving husband,
William Crawford.
"Thank heaven that's the end of this ridiculous connection!" he
said to himself as he dropped the letter into a pillar-box in front of
the club. "My mind is now easy, and I can enjoy myself. I can play

to-night as though I were still a bachelor with no thought of the
morrow. Ah, but I have thought of the morrow! What delightful
thought, too! delightful Hetty."
It was late in the evening when this letter was delivered at
Singleton Terrace. Nothing else came by that post. Although Mrs.
Crawford had often seen her husband's writing, this was the first
letter she had got from him, and she had never before seen her
name and the address of that house in his writing. She did not
recognise the hand, and thinking the letter must be connected with
routine business about the Welford property, she put it on the table
by her bedside unopened. He attended to all such matters.
When the maid brought in her supper she took up the letter again
and turned it over idly in her hands. All at once it struck her that the
writing was familiar, but whose it was she could not guess.
With a smile at her own curiosity, she broke the cover and drew
out the sheet of paper.
She looked at the signature languidly until she read it. Then
hastily, tremulously she scanned the first few lines. When she
gathered their import she uttered a low wailing sob and fell back
insensible on the pillow.

CHAPTER XXXVIII.
WILLIAM CRAWFORD FREE.
When William Crawford had posted his letter to his wife he felt
ten years younger than an hour before. He enjoyed an extraordinary
accession of spirits. The day had grown heavy and cloudy, but to
him it was brighter than the flawless blue of Mediterranean summer.
Richmond and Singleton Terrace were done with for good and all.
There were to be no more private theatricals played for board and
lodgings. Instead of simulating love for an elderly woman, he was at
liberty to make real love to the most charming young girl he had
ever met. His notions of right and wrong were clear and simple:
what he liked was right, what he did not like was wrong. Since he
had come to man's estate he had acted upon the code, and it never
once occurred to him to question it. He did not object to other men
being pious or just or modest; he did not object to their even
preaching a little to him about the merit of these or any other
virtues. All he asked was to be let go his own gait unmolested.
He was now at liberty to take what path he chose and adopt what
sport pleased his humour. He had played for a small fortune and
won. He felt proud of his success, and sorry that the nature of it
forbade him glorying in it. He was aware that the most disreputable
and unprincipled blackleg in the Counter Club would scorn to get
money as he had acquired his. But this did not matter to him. He
was not going to tell any one at the club how he came by the
money; that was an irksome self-restraint imposed upon himself out
of deference to ridiculous conventional ideas. But he had the money
in his pocket--that was the great thing.

As he intended playing all through the night, if the game were
kept up, it was too early to begin at three o'clock in the afternoon.
He should be fagged out before morning if he sat down now. He was
neither so young nor so impetuous that he could not discipline desire
to delay.
All at once he remembered that in abandoning Singleton Terrace
so suddenly he had lost his kit. The value of his baggage was not
very great, and with the sum now in his possession he would not for
three times its value go back to Richmond for it. He had now no
personal belongings but the clothes he stood in and a portmanteau
at Welford. He would go to a tailor and an outfitter and order what
he wanted. That would amuse him and help to kill time. He should
get back to the club about seven, and devote the rest of the evening
and all the night to cards.
He did not go to the tailor with whom he had dealt since he came
to live at Richmond. He wanted to cut himself off from that place as
completely as possible.
At the tailor's he ordered three suits of clothes to be ready in
three days and forwarded to Crawford's House, Crawford Street,
Welford. What he bought at the outfitter's were to be sent to the
tailor's and to accompany the parcel of the latter. He paid in advance
for all. Then he went to another shop, purchased a portmanteau,
and directed it to be delivered at the tailor's, and sent a note with it,
asking him to put the outfitter's parcel and the clothes into it and
send it to the address already given.
Then he bethought him of a dressing-bag, and he bought a
handsome one with silver-mounted bottle and ivory-backed brushes.
The bag, being of leather, reminded him that he had no boots but
those on his feet. So he purchased a couple of pairs and a pair of
slippers, and the slippers put him in mind of a dressing-gown.

He directed all these things to be sent to the tailor's, and wrote to
the tailor to let them all be forwarded at the one time--that is, when
the clothes were finished, in three days.
He enjoyed this shopping greatly. He had never before spent so
much money on himself in one day. It was so pleasant to buy these
articles without worrying about the price, to be in doubt as to
whether he should have a dressing-bag at thirty or thirty-five
pounds, and to decide in favour of the thirty-five-pound one merely
because it had prettier bottles and a greater number of pockets.
When he could think of nothing else which he wanted, he said to
himself, "And now what shall I take Hetty? I must get the very
handsomest present I can light upon."
This set him off calling Hetty up to mind. He looked into the
windows of a dozen jewellers' and shops where fancy articles were
sold. He failed to find an article to his liking. He could not realise
Hetty accepting any of the costly gifts presented to his view. At
length with a sudden start he cried out to himself, "What an idiot I
have been! Of course, she would not accept any of these things
from me now. A few simple flowers from Covent Garden to-morrow
morning on my way to Welford will be the very thing."
It never once occurred to him during the day that the money he
was spending belonged to his wife, and was being laid out in a way
and under conditions not contemplated by her in giving it to him.
When he decided on taking flowers to Hetty, it never once occurred
to him that this would be spending his wife's money to conciliate a
rival of hers, and that twenty-four hours ago he would have bought
these same flowers for his deserted wife.
"Hetty," he said, formulating his theory, "is to be won through her
imagination, not by pelf."
When he got back to the club he reckoned up what he had spent.
It was an agreeable surprise to find that although he had treated

himself with great liberality, all his purchases did not absorb the
hundred-pound note he had changed at the tailor's. He had got a
moderate outfit and a very handsome dressing-case, with cut-glass
bottles silver-mounted, and ivory-backed brushes, for less than one
thirty-second part of the money received from Mr. Brereton that
afternoon. He sat down to an excellent dinner with the conviction
that he had done a fair day's work, and that he was entitled to enjoy
himself for the remainder of the evening, and as far into the
morning as he chose.
The dinner was excellent; his shopping had given him zest for it,
and when he stood up from the table he felt in the most excellent
humour with himself and all the world.
He looked at his watch.
"She has my letter by this time," he said to himself, thinking of his
wife. "If she is not a greater fool than I take her for, she will know
from it that she has seen the last of me."
When he wrote the letter he had no intention of conveying any
such idea to her, but his shopping and thoughts of Hetty had
hardened his heart since then towards his unhappy wife, and now he
wanted to believe that his letter would leave her no loophole of
hope.
"Dr. Loftus said any shock might bring on the end. Perhaps my
letter----" He paused and did not finish the sentence, but began
another: "When a case is hopeless the greatest mercy which can be
shown to the sufferer is, of course, to put an end to the struggle.
She could not have fancied for a moment that I was going to spend
all my life in the sick-room of a woman almost old enough to be my
mother. Anyway, I need not bother my head any more about the
matter. She cannot say that while our married life lasted I was not a
kind and considerate husband. Turn about is fair play, and I am
going to be a little kind and considerate to myself now. I'll put the

past away from my mind. 'Gather we rosebuds while we may' is my
version. Now to lose for the last time."
At the Counter Club there were men every night who did not mind
how far into the morning they sat so long as they were winning.
From the moment Crawford touched the cards until he rose at half-
past six he had lost steadily. Though he had played for higher stakes
than usual, he had been as careful of his game as if he had no more
than a few hundred pounds with him. He had not been reckless. He
had not plunged. Luck had simply been dead against him, and when,
while eating his early breakfast, he counted up the cost, he found he
was close on three hundred pounds the worse for his night's
experience.
Mentally he cursed his bad luck.
"But I deserve no better," he thought. "I told myself that I should
have good luck only when I had come from Welford. The luck I
played with last night was my wife's or my own, and both have been
invariably bad. I shall go to Welford to-day, and play to-night with
Hetty's luck, and win back all I have lost and more besides. And now
to get a bouquet for Hetty--for the loveliest girl in the whole of
England. But the bouquet must not be too splendid. It must be
simple and cheap, or it might do more harm than good."
At Covent Garden he bought some simple blossoms, and had
them tied carelessly together.
"She will not value them for what they cost, but for my
remembering her."
He was full of confidence in his power to fascinate and win. It
never for a moment occurred to him that Hetty might not care for
him or his memory of her. The notion of a rival had never entered
his head, and if any one had suggested such a thing he would have
laughed the consideration of it to scorn. He admired Hetty intensely,
and he meant to succeed, and succeed he would.

Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade
Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!
textbookfull.com

Robot Systems for Rail Transit Applications 1st Edition Hui Liu

More Related Content

Similar to Robot Systems for Rail Transit Applications 1st Edition Hui Liu (20)

Recently uploaded (20)

Robot Systems for Rail Transit Applications 1st Edition Hui Liu