MINIMIZATION OF METABOLIC COST OF MUSCLES BASED ON HUMAN EXOSKELETON MODELING: A SIMULATION

International Journal of Biomedical Engineering and Science (IJBES), Vol. 3, No. 4, October 2016
DOI: 10.5121/ijbes.2016.3401 01
MINIMIZATION OF METABOLIC COST OF
MUSCLES BASED ON HUMAN EXOSKELETON
MODELING: A SIMULATION
Harita Baskar and Sri Madhava Raja Nadaradjane
Department of Electronics and Instrumentation, St. Joseph’s College of Engineering,
Anna University, Chennai, India
ABSTRACT
In this work, movement of the exoskeleton wearer and the metabolic energy changes with the assisted
devices using OpenSim platform has been attempted. Two musculoskeletal models, one with torsional ankle
spring and the other with bi-articular path spring are subjected to forward dynamic simulation.The
changes in the metabolic rate of the lower extremity muscles before and after the addition of the assistive
devices were tested. The results about the effect of these external devices on individual muscles of the lower
muscle group were analysed which provided effective results.
KEYWORDS
Human Exoskeleton, Metabolic Cost, Gastrocnemius muscle, Soleus muscle & Illiopsoas muscle
1. INTRODUCTION
An exoskeleton is a particular kind of mechanical device designed to be worn by an operator that
help in assisting limb movement and the execution of a motor task [1]. These exoskeletons are
widely used for various orthopaedic disorders, different levels of paralysis, military applications
and in rehabilitation centres [2]. The main target of this technology is to augment the human body
and its capabilities[3].
Exoskeletons have the potential to reduce the metabolic cost of walking, loaded walking as well
as running. Figure1 shows the main components of an exoskeleton on a loaded subject. It includes
hip springs, adduction springs and an ankle spring. These spring components help in reducing the
metabolic energy of the muscles. The knee damper helps in limiting the range of motion [4]. The
reduction in metabolic cost decreases the possibilities of injury and increases the load carrying
capacity [5]. Researchers have attempted to achieve this by developing both active and passive
exoskeletons. Kazerooni, Racine, et. al. developed a lower extremity exoskeleton for load
carriage with actuated hip, knee and ankle joints which described an autonomous model
[6].Walsh et. al. designed an quasi passive leg exoskeleton with only ankle and hip springs with a
knee variable damper [7]. These devices try to assist human gait but there are certain challenges
associated with such devices. It is difficult to analyse the effectiveness of the device and also
cannot predict how external actuation assists muscles during loaded walking, Simulation can help
develop the wearable device, as it can give an idea on how the device helps muscles and how the
metabolic cost changes during loaded walking. Muscle driven simulation allows the calculation of
muscle forces, residual forces, fiber length and other parameters that cannot be easily measured. It
is very important to measure these neuromuscular quantities, as they are responsible for the
production of movement [8]. It also makes it possible to measure the metabolic cost of different
muscle groups during the gait cycle.

International Journal of Biomedical Engineering and Scie
In order to achieve a simulation based design there are certain tools such as AnyBody modelling
system [9], SimTK[10], Virtual Interactive Musculoskeletal System (VIMS)[
(Software for Interactive Musculoskeletal Modeling)[
the effect of elastic ankle exoskeleton
tendon dynamics with a generic musculoskeletal model
the effectiveness of powered exoskeleton using computer simulation that aimed to decrease the
metabolic energy consumption for upper arm reaching
Figure 1.Components sketch of the main components of the exoskeleton
This paper focuses on the simulation of movement of the exoskeleton wearer using Open
platform and the metabolic energy c
This exoskeleton is based on the spring like action of leg
human stance during walking, it is seen that there exist a linear relationship among the dorsi
flexion and plantar-flexion stages of the stance Therefore, the metabolic
reduced by replacing it with a torsional ankle spring.
the spring is able to minimize the energy expenditure during walking by storing energy during the
progression stage and allowing free movement in rest of the gait cycle [
simulation is performed for two musculoskeletal models, one with torsional ankle spring and the
other with bi-articular path spring. T
muscles after and before the addition of the assistive devices
effects of these external devices on Gastrocnemius muscle, Soleus muscle and Illiopsoas muscle
were also analysed.
2. METHODOLOGY
In this paper, two three-dimensional musculoskeletal models with 10 degrees of freedom with 18
muscle tendon paths with actuators were created using OpenSim. Both the model
male of 75 (Kg) mass and 1.8(m) height. The
for the forward dynamic simulation.
and foot segments [17]. The two models are subjected to three processes before the addition of
the assistive devices which includes scaling, inverse kinematics and computes muscle control.
These three processes were used to create simulations of the
1.3(m/s). Then a torsional ankle
the other model.
International Journal of Biomedical Engineering and Science (IJBES), Vol. 3, No. 4, October
], Virtual Interactive Musculoskeletal System (VIMS)[
(Software for Interactive Musculoskeletal Modeling)[12] and OpenSim[13]. Dominic et al studied
the effect of elastic ankle exoskeleton on bilateral hopping by utilizing simulation of muscle
with a generic musculoskeletal model[14]. Recently, Yoshiaki et al evaluated
metabolic energy consumption for upper arm reaching movement [15].
1.Components sketch of the main components of the exoskeleton
This paper focuses on the simulation of movement of the exoskeleton wearer using Open
platform and the metabolic energy changes associated with the lower extremity muscle
eleton is based on the spring like action of leg and from the moment angle analysis of
human stance during walking, it is seen that there exist a linear relationship among the dorsi
flexion stages of the stance Therefore, the metabolic cost of the ankle joint is
reduced by replacing it with a torsional ankle spring. In conjunction with a damping mechanism,
progression stage and allowing free movement in rest of the gait cycle [16]. Forward dynamic
d for two musculoskeletal models, one with torsional ankle spring and the
th spring. The changes in the total metabolic cost of the lower extremity
the addition of the assistive devices were examined. More
devices on Gastrocnemius muscle, Soleus muscle and Illiopsoas muscle
dimensional musculoskeletal models with 10 degrees of freedom with 18
with actuators were created using OpenSim. Both the model
75 (Kg) mass and 1.8(m) height. The dynamic musculoskeletal model is used as the input
simulation.. The model consist of head, trunk, pelvis,both femur,
used to create simulations of the subject running at a speed of
ankle spring is attached to one model and a bi-articular path spring to
nce (IJBES), Vol. 3, No. 4, October 2016
2
], Virtual Interactive Musculoskeletal System (VIMS)[11],SIMM
Dominic et al studied
by utilizing simulation of muscle
Recently, Yoshiaki et al evaluated
This paper focuses on the simulation of movement of the exoskeleton wearer using Open Sim
hanges associated with the lower extremity muscle group.
and from the moment angle analysis of
human stance during walking, it is seen that there exist a linear relationship among the dorsi-
cost of the ankle joint is
In conjunction with a damping mechanism,
Forward dynamic
d for two musculoskeletal models, one with torsional ankle spring and the
of the lower extremity
Moreover, the
devices on Gastrocnemius muscle, Soleus muscle and Illiopsoas muscle
dimensional musculoskeletal models with 10 degrees of freedom with 18
with actuators were created using OpenSim. Both the models represent a
musculoskeletal model is used as the input
. The model consist of head, trunk, pelvis,both femur, tibia
ubject running at a speed of
articular path spring to

3
Figure 2 Flowchart for dynamic simulation
2.1 Scaling
The initial step to perform forward dynamic simulation is to scale a generic model to fit a
particular data. It is scaled to match the required size and weight of the model. By using motion
capture equipment, the marker locations are identified and this represents the experimental
markers. The unscaled model already has a set of virtual markers around the same location as the
experimental markers. The dimensions of each segment in the model are scaled by moving the
virtual markers such that they coincide with the experimental marker locations [18].
2.2 Inverse kinematics
Inverse kinematics is performed to determine the generalized coordinates which represents the
joint angles and translations of the model and to minimize the coordinate and marker errors .The
IK tool in OpenSim ensures that the model is placed in a pose that best matches the experimental
marker locations in each time frame. This is achieved through solving the weighted least square
problem and the solution is mathematically expressed as
∑ || − || + ∑ −

!(1)
Where q is the vector of generalized coordinates, xi
exp
is the experimental position of marker i,
xi(q) is the position of the corresponding marker on the model (which depends on the coordinate
values), qj
exp
is the experimental value for coordinate j[19]. The weighted square error that has to
be minimized is given as

4
" #$%&' &%%(% = ∑ * − *
+
+ ∑ ,
-+
− , +
(2)
Where * and *
+
are the three dimensional position of the .th marker for the model,
, and , +
are the values of the /th joint angles for the model and and are the
weight factors [13].
2.3 Computed muscle control
Computed muscle control (CMC) is used to generate a set of muscle excitation that drives the
generalized coordinates to closely track the desired kinematics. Before using the CMC tool, the
initial condition values which include the values of generalized velocities, generalized
coordinates and muscle activation length have to be applied for the first 0.03 seconds of the
desired interval. The forward dynamic simulation is attained by the CMC tool with the
combination of static optimization and a proportional derivative control.Static optimization is
used as an actuator controller to achieve the desired accelerations and helps in eliminating any
noise in the data when a there is a sudden difference in acceleration during the gait cycle. The
desired accelerations are computed based on the proportional-derivative control law that is
written as
0 1 + ∆1 = 0 1 + ∆1 + 34 [ 6 1 − 1 ] +6 3 [ 1 − 1 ] (3)
Where84 and 8 are the feedback gains on the velocity and position errors, 0 represents the
desired accelerations whereas and represent the model coordinates and experimentally
derived coordinates respectively[19].CMC is carried out until the desired movement is achieved.
Figure 3 Model without assistance
2.4 Simulation with assistive devices
After the completion of the above processes, forward dynamic simulation is performed with two
assistive devices: torsional ankle spring and bi- articular path spring. The torsional ankle spring
is modelled at the right ankle joint and the bi-articular path spring acts long the right femur and
foot segment. Both the springs act as a force element and functions as a spring and damper with a
specified stiffness value. The stiffness value for the ankle spring and bi-articular path spring is set
as 10 (Nm) and 10000 (Nm) respectively and damping ratio is 0.01. The two force elements are
assigned as a Coordinate Limit Force and this limits the range of motion when the dorsiflexion
angle exceeds a specified angle(set as 5 degrees). Thus two simulations are executed with each of
the springs modelled individually.

5
(a) (b)
Figure 3. (a) Model With Torsional Ankle Spring, Blue Markers Identify The Connection Points (b) Model
With Bi-Articular Path Spring, Green Marker Identifies The Connection Points.
3. RESULTS AND DISCUSSIONS
3.1 Without assistance
Computed muscle control (CMC) was used to generate muscle driven simulation for model
without any assistance. After the application of computed muscle control, the total metabolic
energy was calculatedusing the metabolic cost calculator at various time intervals. Figure 4
illustrates the total metabolic energy, which ranges from 500J-2000J. The maximum value of
metabolic energy is 1852.54 (Joules) at time 1.35 seconds.
Figure 4. Total Metabolic Energy Vs Time For Unassisted Model

6
3.2 With torsional ankle spring
A torsional ankle spring is added to the model by using the OpenSim tool. This aids in reducing
the total metabolic energy of the lower extremity. The blue curve in fig 2 indicates the total
metabolic energy after adding the torsional ankle spring, which shows a 30% decrease in the total
energy when compared to the unassisted model. The maximum value of total energy is
1255.18(Joules) at time 0.7 (sec).
Figure 5. Comparison of Total Metabolic Energy vs Time For Unassisted
3.3 With Bi-articular path spring
This spring reduces the metabolic energy of the muscles along the femur and the foot segments. It
significantly reduces the metabolic energy of the gastrocnemius muscle, illiopsoas muscle and
soleus muscle. Figure 6 indicates that there is a 30% decrease in the total metabolic energy when
compared with the metabolic energy generated by the unassisted model. The maximum value
total metabolic energy is 1252.14(Joules) at 0.7(seconds) of the gait cycle.
Figure 6. Total metabolic energy vs time for a model with bi-articular path spring

7
Table 1 Comparison of total metabolic energy of unassisted model with assisted models with respect to
time
S.No Time
(sec)
Total metabolic energy(Joules)
Unassisted
model
Model with
Torsional
ankle spring
Model with
Bi-articular
path spring
1. 0.7 1752.10 1255.18 1252.14
2. 0.8 1101.29 772.19 662.73
3. 0.9 592.89 656.71 543.92
4. 1.0 231.42 110.32 201.15
5. 1.1 361.14 301.11 431.29
6. 1.2 701.91 737.81 744.56
7. 1.3 553.12 624.43 251.03
8. 1.35 1852.54 901.73 826.15
9. 1.4 761.22 770.10 669.45
3.4 Effect of Bi-articular path spring on the lower extremity muscles:
The spring helps in reducing the metabolic energy of the Gastrocnemius muscle, Soleus muscle
and Illiopsoas muscle. By using the OpenSim tool, the total metabolic energy with respect to time
was obtained. From the simulation plots, the average of the total metabolic energy is calculated
for each muscle group before and after the addition of the Bi-articular path spring. After the
comparison of the results, it is seen that there is a substantial decrease in the metabolic energy
after the addition of the Bi-articular path spring. The average value of metabolic energy for
Gastrocnemius muscle is 95.487(Joules), for Soleus muscle it is 46.604(Joules) and for Illiopsoas
muscle it is 50.743(Joules).
(a)

8
(b)
(c)
Figure 7. (a) Gastrocnemius muscle metabolic rate (b) Illiopsaos muscle metabolic rate
(c) Soleus muscle metabolic rate
Table 2 Average value of metabolic energy with and without Bi-articular path spring
Muscle group
Metabolic energy (Joules)
Without bi-articular
spring
With bi-articular
spring
Gastrocnemius
muscle
102.71 67.48
Soleus
Muscle
98.16 46.60
Illiopsoas
Muscle
88.760 50.74
4. CONCLUSION
In this paper, the modelling of an exoskeleton with two assisted devices and the simulation of the
exoskeleton wearer has been executed. The total metabolic energy for different muscle groups
were compared with and without the assisted devices. It was examined, thatthe metabolic energy
of each muscle has reduced with the addition of a torsional ankle spring and bi-articular path
spring. For future work, these simulations can aid in evaluating the performance of powered
prosthetic legs and can provide inputs for alternate prosthesis design. Furthermore, the analysis of
metabolic cost of the lower extremity muscles (Gastrocnemius and soleus muscle) can forma
foundation for minimizing the effects of osteoporosis and medial calf injuries.

9
REFERENCES
[1] A. Moubarak, M. Tu Pham, T. Pajdla and T. Redarce, (2009) , "Design and Modeling of an Upper
Extremity Exoskeleton", 11th International Congress of the IUPESM : medical physics and biomedical
engineering world congress 2009, Springer.pp.476-479.
[2] J. C. Perry, J. Rosen and S. Burns, (2007),"Upper-Limb Powered Exoskeleton Design",IEEE/ASME
TRANSACTIONS ON MECHATRONICS, vol. 12, no. 4.
[3] J. E. Pratt, B. T. Krupp and C. J. Morse, (2004),"The RoboKnee: An Exoskeleton for Enhancing
Strength and Endurance During Walking", Proceedings of the 2004 IEEE International conference on
Robotics and Automation.
[4] C. James Walsh, D. Paluska, K. Pasch, W. Grand, A. Valiente and H. Herr, (2006),"Development of a
lightweight, underactuated exoskeleton for load-carrying augmentation", Proceedings of the 2006
IEEE International Conference on Robotics and Automation.
[5] L. M Mooney, E. J Rouse and H. M Her, (2014),"Autonomous exoskeleton reduces metabolic cost of
human walking during load carriage", Journal of NeuroEngineering and Rehabilitation, pp. 11:80.
[6] A. B. Zoss and A. Chu, (2005), "On the Mechanical Design of the Berkeley Lower Extremity
Exoskeleton (BLEEX)", IEEE/RSJ International Conference on Intelligent Robot and Systems.
[7] C. James Walsh, K. Endo and H. Herr, (2007), "Design of a Quasi-Passive Parallel Leg Exoskeleton to
Augment Load Carrying for Walking", International Journal of Humanoid Robotics, vol. 4, no. 3, pp.
487–506.
[8] P. Agarwal, M. S. Narayanan, L. Lee, F. Mendel and V. N. Krovi, (2010), "Simulation-Based Design
of Exoskeletons Using Musculoskeletal Analysis", International Design Engineering Technical
Conferences and Computers and Information in Engineering Conference.
[9] A.Technology, "AnyBody Technology", Anybodytech.com, (2016), [Online]. Available:
http://www.anybodytech.com [Accessed: 03- Jul- 2016].
[10] "SimTK: Welcome", Simtk.org, (2016 )[Online]. Available: https://simtk.org [Accessed: 03- Jul-
2016].
[11] Chao, Y., et al., (2007), "Virtual Interactive Musculoskeletal System (VIMS) in OrthopedicResearch,
Education and Clinical Patient Care," Journal of Orthopaedic Surgery and Research, pp. 1-19.
[12] "SIMM by MusculoGraphics, Inc.", Musculographics.com, (2016), [Online]. Available:
http://www.musculographics.com [Accessed: 03- Jul- 2016]
[13] Delp SL, Anderson FC, Arnold AS, Loan P, Habib A, John CT, Guendelman E, Thelen DG,
(2007)“OpenSim: Open-source Software to Create and AnalyzeDynamic Simulations of Movement”,
IEEE Transactions on Biomedical Engineering .
[14] F. GS, "Linking the mechanics and energetics of hopping with elastic ankle exoskeletons. - PubMed -
NCBI", Ncbi.nlm.nih.gov, (2016), [Online].Available:
http://www.ncbi.nlm.nih.gov/pubmed/23065760. [Accessed: 02- Jul- 2016].
[15] Y.Tanjal, T.Nanjwa, Y.Takahashi, M. Kawal, (2016), “Evaluation of power assist System By
Computer Simulation”, JCII,Vol.20,No.3,pp.447-483.
[16] Kamran Shamaei, Massimo Cenciarini, and Aaron M. Dollar, (2011), “On the Mechanics of the Ankle
in the Stance Phase of the Gait”, 33rd Annual International Conference of the IEEE EMBS.
[17] Ajay Seth, Darryl Thelen, Frank C. Anderson, Scott L. Delp, (2005), “Model of trunk, pelvis and leg
segments,10 degrees of freedom and 18 muscles”, Annals of Biomedical Engineering, vol.33, pp 829-
840.
[18] "How Scaling Works - OpenSim Documentation -", Simtk-confluence.stanford.edu, 2016. [Online].
Available:http://simtk-
confluence.stanford.edu:8080/display/OpenSim/How+Scaling+Works#HowScalingWorks-
Scaling.[Accessed: 04- Jul- 2016].
[19] M. Margarida, (2013) "A multibody approach to the contact dynamics : a knee joint
application", Hdl.handle.net. [Online]. Available: http://hdl.handle.net/1822/24564. [Accessed: 04-
Jun- 2016]

MINIMIZATION OF METABOLIC COST OF MUSCLES BASED ON HUMAN EXOSKELETON MODELING: A SIMULATION

More Related Content

What's hot (20)

Viewers also liked (16)

Similar to MINIMIZATION OF METABOLIC COST OF MUSCLES BASED ON HUMAN EXOSKELETON MODELING: A SIMULATION (20)

Recently uploaded (20)

MINIMIZATION OF METABOLIC COST OF MUSCLES BASED ON HUMAN EXOSKELETON MODELING: A SIMULATION