Available via license: CC BY 4.0
Content may be subject to copyright.
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
1
Abstract—Objective: A unilateral, lightweight powered hip
exoskeleton has been shown to improve walking economy in
individuals with above-knee amputations. However, the
mechanism responsible for this improvement is unknown. In this
study we assess the biomechanics of individuals with above-knee
amputations walking with and without a unilateral, lightweight
powered hip exoskeleton. We hypothesize that assisting the
residual limb will reduce the net residual hip energy.
Methods: Eight individuals with above-knee amputations
walked on a treadmill at 1 m/s with and without a unilateral
powered hip exoskeleton. Flexion/extension assistance was
provided to the residual hip. Motion capture and inverse dynamic
analysis were performed to assess gait kinematics, kinetics, center
of mass, and center of pressure.
Results: The net energy at the residual hip decreased from
0.05±0.04 J/kg without the exoskeleton to -0.01±0.05 J/kg with the
exoskeleton (p = 0.026). The cumulative positive energy of the
residual hip decreased on average by 18.2% with 95% confidence
intervals (CI) (0.20 J/kg, 0.24 J/kg) and (0.16 J/kg, 0.20 J/kg)
without and with the exoskeleton, respectively. During stance, the
hip extension torque of the residual limb decreased on average by
37.5%, 95% CI (0.28 Nm/kg, 0.36 Nm/kg), (0.17 Nm/kg, 0.23
Nm/kg) without and with the exoskeleton, respectively.
Conclusion: Powered hip exoskeleton assistance significantly
reduced the net residual hip energy, with concentric energy being
the main contributor to this change. We believe that the reduction
in residual hip extension torque during early stance is the main
contributor to this reduction.
Significance: This analysis shows that by assisting the residual
hip, the exoskeleton significantly decreased the net hip energy
produced by the residual limb, which may explain the
improvements in walking economy previously observed.
Index Terms—Powered Hip Exoskeleton, Robotics, Biomimetic
& Bio-Inspired Robotics, Transfemoral Amputation, Prosthetics,
Assistive/Rehabilitation Robotics, Biomechanics
This paper was submitted for review in December 2021. This work was
supported by the U.S. Department of Defense under Grant # W81XWH-16-1-
0701 and by the National Science Foundation under Award # 2046287.
All authors are currently affiliated with the University of Utah. a Indicates
authors affiliated with the Department of Mechanical Engineering, Robotics
I. INTRODUCTION
onventional knee and ankle prostheses are energetically
passive devices that cannot provide biomechanically
accurate torque, power, and movements during ambulation [1].
Individuals with above-knee amputations compensate for the
limited functionality of their passive prostheses with complex
compensatory strategies involving their sound leg, their
residual limb above the amputation level, their pelvis, and their
upper body [2], [3]. Clinical research has shown that these
compensatory gait strategies have negative effects on walking
balance, stability, speed, and efficiency [4], limiting the
movement ability and quality of life of individuals with leg
amputations. Moreover, the compensatory gait strategies used
with lower-limb prostheses have been frequently linked to
debilitating secondary health issues such as osteoarthritis [5]
and lower-back pain [6]. These studies motivate the
development of new assistive technologies and interventions
for individuals with lower-limb amputations.
Clinical gait analysis highlights the functional limitations of
existing prostheses and the compensatory strategies used by
individuals with above-knee amputations during walking.
Inverse dynamics analysis of nonamputee gait shows that the
biological ankle provides net-positive energy [7], [8]. However,
most conventional prostheses are passive. Therefore, they
cannot provide energy injection, which is a critical function of
biological legs. Individuals with above-knee amputations
compensate for the loss of biological ankle energy on the
prosthesis side using different strategies. One strategy is to
increase the effort in their residual, prosthesis side hip [9]. This
residual hip compensation strategy is particularly hard for
individuals with above-knee amputations because the residual
limb muscles are rearranged during amputation surgery in a
non-physiological manner [10], which reduces the strength and
range of motion of the residual hip joint [11]. Individuals with
Center. b Indicates authors affiliated with the Department of Biomedical
Engineering. c Indicates authors affiliated with the Department of Physical
Therapy and Athletic Training.
Powered Hip Exoskeleton Reduces Residual
Hip Effort without Affecting Kinematics and
Balance in Individuals with Above-Knee
Amputations During Walking
Marshall K. Ishmaela, Andrew Gunnella, Kai Pruyna,b, Suzi Crevelinga, Grace Hunta, Sarah Hooda,
Dante Archangelia, K. Bo Foremana,c, Tommaso Lenzia
C
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
2
above-knee amputations also compensate for the lack of energy
from their prostheses by increasing the effort in their sound hip
and ankle [12]. Due to these compensatory movements, the
pelvic motion increases by a factor of two [13], [14]. Although
these compensatory movements are observed in many subjects
[15], trunk lean and atrophied hip muscles due to amputation
surgery may actually result in reduced prosthesis side hip effort
[16]. Moreover, knee prosthesis selection has a noticeable
effect on hip kinematics [17] and walking stability [18].
Similarly, socket selection and the residual limb length affect
the kinematics of the residual hip joint [13]. These studies
provide a mechanistic explanation for the asymmetric gait
pattern observed in individuals with above-knee amputations,
motivating the development of more advanced assistive
technologies that can overcome the fundamental limitations of
existing leg prostheses [19], [20].
Powered prostheses have been proposed to more closely
replicate the biomechanical function of the missing biological
leg. For example, powered prostheses can inject net-positive
energy into the gait cycle with battery-powered servomotors
which can imitate the energy-injection function of the
biological ankle [21], [22]. Unfortunately, existing powered
prostheses are substantially heavier than their passive
counterparts. Biomechanical studies have shown that additional
mass placed on body segments increases the effort required to
walk, and that the additional energetic exertion is proportional
to the distance of the additional mass from the whole-body
center of mass. For example, adding mass at the ankle increases
metabolic cost four times more than at the trunk [23].
Moreover, increased prosthesis mass correlates with increased
energy expenditure and increased gait asymmetry [24]. For
example, matching the prosthesis weight to the weight of the
missing biological limb resulted in a 12% increase in the
metabolic cost of walking [25], and increasing the inertia of the
prosthesis exacerbated stance and swing time asymmetries [26].
Despite recent efforts [27], [28], reducing the weight of
powered prostheses without deteriorating performance is an
open challenge. Therefore, the potential benefits of powered
prostheses are offset by the increased distal mass of these
devices.
An alternative approach to improve amputee walking
economy is assisting the residual or sound hips with a
motorized orthosis or a powered hip exoskeleton [29]–[33]. A
powered hip exoskeleton can be made lightweight because the
onboard motors only need to provide a fraction of the torque
generated by the biological hip. Moreover, the mass of a
powered hip exoskeleton is located close to the trunk,
minimizing the negative effects on user’s gait mechanics and
energetics [23]. Perhaps because of their small and proximal
mass, powered hip exoskeletons have recently succeeded in
significantly reducing the metabolic cost of walking in
individuals with above-knee amputations [29], [33], which has
never been shown with powered prostheses. Powered hip
exoskeletons have also been effective gait training devices [30]
and stumble recovery aids in individuals with above-knee
amputations [31]. These positive clinical outcomes suggest that
powered hip exoskeletons are a viable tool to improve gait in
individuals with above-knee amputations. However, the effect
of the assistance provided by a hip exoskeleton on amputee gait
mechanics is still unknown. Thus, there is no mechanistic
explanation for the improved metabolic cost of walking that has
been shown with powered hip exoskeletons.
In this paper, we assess the effect of residual limb assistance
provided by a unilateral, lightweight powered hip exoskeleton
on gait kinematics and kinetics of individuals with above-knee
Table I.
Participant Demographics
Subject
Age
[yrs]
Weight
[kg]
Height
[m]
Sex
Years since
Amputation
Knee
Prosthesis
Ankle
Prosthesis
Amputation
Side
Socket
Suspension
S1
73
79.5
1.65
M
5
C-Leg
Triton
L
Lanyard
S2
52
109.1
1.73
M
4
Genium X3
Triton
R
Lanyard
S3
37
100.5
1.80
M
9
C-Leg
All Pro
L
Suction
S4
28
65.1
1.78
M
7
Plie
All Pro
R
Suction
S5
31
77.3
1.80
M
3
KX06
Hi Pro
L
Suction
S6
31
59.1
1.60
F
12
Plie
All Pro
L
Lanyard
S7
51
102.3
1.91
M
9
C-Leg
Triton
L
Pin Lock
S8
39
90.5
1.91
M
35
Plie
Soleus
L
Suction
Figure 1. Experimental setup. (a) A subject walks on the instrumented
treadmill with reflective markers without the exoskeleton (b) on the
instrumented treadmill with reflective markers with the exoskeleton (c) red
circles indicate markers on the rigid exoskeleton frame, and the blue
rectangle highlights markers on the prismatic passive degree of freedom.
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
3
amputations. We hypothesize that assisting the residual limb
with a unilateral, lightweight powered hip exoskeleton will
decrease the torque generated by the residual limb muscles at
the hip joint, reducing the residual limb energy. We will test this
hypothesis by performing motion capture and inverse dynamic
analysis while eight participants with unilateral above-knee
amputations walk with and without an autonomous, lightweight
powered hip exoskeleton assisting their residual limb. We will
use net hip energy at the residual limb as the primary outcome.
We will also perform secondary analyses on kinematics, gait
timing, balance, and stability to assess the effect of asymmetric
assistance on compensatory behaviors, as well as to assess
changes associated with wearing the assistive exoskeleton.
By providing new insights into how above-knee amputees
adapt their biomechanics to exoskeleton assistance, this paper
contributes new knowledge related to the residual hip-
assistance approach. This knowledge may suggest new ways to
improve clinical outcomes for individuals with above-knee
amputations.
II. METHODS
A. Unilateral, lightweight powered hip exoskeleton
The unilateral, lightweight powered hip exoskeleton used for
this study (Figure 1 (a)) was previously presented and validated
in healthy and amputee subjects [32]. This exoskeleton provides
flexion and extension torque to the user’s residual limb while
allowing for passive abduction and adduction. The exoskeleton
connects to the user’s socket and includes passive degrees of
freedom to reduce the spurious forces and torques on the user’s
leg [36], improving comfort [37]. The exoskeleton’s embedded
control system runs an assistive controller based on adaptive
frequency oscillators [38]–[40]. Key parameters of the assistive
controller such as the level and timing of the assistance in
flexion and extension were tuned for each participant before the
start of the experiment. Tuning was based on feedback from the
subjects as well as the experimenter’s experience.
B. Experimental Protocol
We recruited eight participants with unilateral above-knee
amputations to participate in this study. Inclusion criteria
included age between 18 and 85 years, at least 1 year post
amputation surgery, daily use of their prescribed prosthesis, and
ability to walk on a treadmill without using handrails. Exclusion
criteria included serious comorbidities (including
musculoskeletal, cardiac, neuromuscular, skin, or vascular
conditions) and inability to communicate or be understood by
investigators. All participants had prior experience walking on
a treadmill and were considered to be Medicare classification
level K3. Information regarding the recruited participants can
be found in Table I. The Institutional Review Board at the
University of Utah (IRB00099066, approved 07/15/2017) and
the U.S. Human Research Protection Office (HRPO) of the U.S.
Army Medical Research and Development Command (HSRRB
log number: A-19840) approved the study protocol. All
participants provided informed consent to participate in the
study and to be photographed and video-taped for publication.
Participants wore tight fitting clothing with reflective,
cutaneous markers placed on anatomical bony landmarks
(Figure 1). Markers were placed following a modified Plug-in-
Gait model with redundant markers placed on the hip, thighs,
shins, and feet for more reliable tracking [15]. Additional
markers were placed on the pelvis orthosis, which connected
the exoskeleton to the user’s waist. As shown in Figure 1 (a),
eight markers were placed on the unilateral hip exoskeleton –
four to track the prismatic passive degree of freedom and four
to track the position and orientation of the actuator relative to
the residual limb. Participants wore a harness while walking on
the treadmill in case of an adverse walking event.
Calibration of the motion capture model was performed
using three routines. Each subject completed a static calibration
trial, a functional joint center calibration trial, and a functional
dynamic capture. These calibration routines were repeated
twice, once without the exoskeleton and once with the
exoskeleton. The static calibration created a subject-specific
motion capture model. The functional joint center calibration
located the centers of rotation for the hip using the Symmetric
Center of Rotation Estimation (SCoRE) [34] and knee joint
axes using SCoRE and Symmetrical Axis of Rotation Analysis
(SARA) [35]. The functional dynamic capture improved the
ability of the software to build and label the model for each
additional capture. Upon completion of the calibration trials,
subjects walked for six minutes on a split-belt treadmill (Bertec,
Ohio, USA) at 1 ms-1 with instrumented force plates. During the
first capture, subjects walked without wearing the exoskeleton,
but while wearing the pelvis orthosis to ensure hip markers
remained the same between no exoskeleton and exoskeleton
conditions. Following the first six-minute capture, subjects
donned the unilateral exoskeleton and performed the second set
of calibration routines. Subjects walked on the treadmill with
the exoskeleton for three six-minute captures to allow subjects
to acclimate to the exoskeleton assistance, as in our previous
metabolic study [33]. The final fifty strides the no exoskeleton
trial and the final fifty strides from the third exoskeleton trial
were used for inverse dynamic analysis, as in our previous work
[15]. Marker trajectories were sampled at 200 Hz, and ground
reaction forces were sampled at 1000 Hz. All data were
synchronized during the capture.
Table II.
95% Confidence Intervals for Residual Hip Joint Energy, Power, and
Torque
Exo No Exo
H1
Energy (J/kg) (0.03, 0.05) (0.06, 0.08)
Power (W/kg) (0.19, 0.33) (0.34, 0.46)
H2
Energy (J/kg) (-0.15, - 0.19) (-0.13, -0.15)
Power (W/kg) (-1.09, -1.27) (-1.28, -1.46)
H3
Energy (J/kg) (0.10, 0.12) (0.12, 0.14)
Power (W/kg) (1.26, 1.50) (1.33, 1.53)
Peak Extension Torque After
Heel Strike [Nm/kg]
(0.17, 0.23) (0.28, 0.36)
Peak Flexion Torque [Nm/kg] (0.86, 1.22) (0.90, 1.22)
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
4
C. Data Analysis
Data were analyzed offline using Nexus (Vicon Motion
Systems, Ltd., Oxford, UK), Visual 3D (C-Motion, Maryland
USA), and MATLAB (MathWorks, Inc. Massachusetts, USA).
Marker trajectories and force plate measurements were digitally
lowpass filtered using bidirectional, fourth order Butterworth
filters at 6 Hz and 15 Hz, respectively. Cutoff frequencies were
determined by residual analysis and visual inspection [41].
Ankle joint centers were estimated by markers at the ankle. The
SCoRE model was used to calculate the hip kinematics. Both
the SCoRE and SARA models were used to calculate the knee
kinematics. A variation of the Charnwood Dynamics Model
(CODA) pelvis, namely the V3D Composite Pelvis, was used
for all hip kinematic analyses. All inverse dynamic calculations
were computed using Visual 3D. All data were exported to
MATLAB for subsequent analysis.
As is commonly done in the powered exoskeleton field [42]–
[44], we calculated the biological torque produced by the
residual limb of the participants by assuming that the
exoskeleton perfectly transfers its desired assistance to the user.
Thus, the “biological residual hip torque” was calculated by
subtracting the assistive torque provided by the exoskeleton
from the “total residual hip torque” calculated through inverse
dynamics.
After stride-time normalization, we calculated the average
kinematic and kinetic profiles first for each subject across
strides, and then calculated the average across the resultant
subject means. We also calculated the average of maximum
values for each subject across strides, and then calculated the
average across the resultant subject means.
Energy was calculated as the time integral of power during
each stride. Energetic analyses at the residual hip were divided
into subphases as identified in [11]. These phases are identified
as H1, H2, and H3, where H1 corresponds to concentric
extension power after heel strike, H2 corresponds to eccentric
flexion power prior to toe-off, and H3 corresponds to concentric
flexion power at toe-off.
To assess changes in center of mass motion, we used the
extrapolated center of mass, which relates the lateral center of
mass position and the lateral center of mass velocity divided by
the subject’s eigenfrequency (defined by the inverted pendulum
model). This method has been previously developed and tested
on amputee and nonamputee populations [3], [45]. The
extrapolated center of mass is useful for assessing stability in
dynamic situations [46]. For walking to be considered stable,
the extrapolated center of mass should fall between the centers
Figure 2. (a) Net residual hip energy (i.e., prosthesis side) averaged across all strides and subjects (bar heights) with individual subject means (circles).Asterisk
* indicates statistically significant difference(p<0.05) between the “No Exo” and “With Exo” conditions, (b) residual hip energy in concentric and eccentric
sub-stride phases, averaged across all strides and subjects (bar heights) with individual subject means (circles), (c) residual hip torque, averaged across all
strides and subjects (lines) and standard error of the mean (shading) and (d) residual hip power.). Gray indicates the no exoskeleton condition while red
indicates the exoskeleton condition. Bar plots in (b) correspond to the integral of the power phases called out in plot (d).
No Exo
With Exo
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
Net Residual Hip Energy [J/kg]
0
0.05
0.1
0.15
0.2
Concentric
H1 Energy [J/kg]
0
0.1
0.2
0.3
Eccentric
H2 Energy [J/kg]
No Exo
With Exo
0
0.05
0.1
0.15
0.2
Concentric
H3 Energy [J/kg]
0 20 40 60 80 100
-0.5
0
0.5
1
Residual Hip
Torque [Nm/kg]
H1
H2
H3
0 20 40 60 80 100
Stride [%]
-1
-0.5
0
0.5
1
Residual Hip
Power [W/kg]
(a) (b) (d)
(c)
*
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
5
of pressure for the left and right legs. Accordingly, one measure
of stability can be calculated by the minimum distance between
the average center of pressure of one limb to the peak value of
the extrapolated center of mass within a stride.
D. Statistical Analysis
Statistical analysis was performed only on the primary
outcome of the study—the net residual hip energy. Specifically,
we performed a paired, two-tailed t-test using MATLAB after
verifying normality and homogeneity of variance. We reported
95% confidence intervals (CI) for the no exoskeleton vs
exoskeleton conditions for all secondary outcomes, including
spatiotemporal, kinematic, and kinetic parameters. These
secondary outcomes were analyzed to assess the effect of
asymmetric assistance on compensatory behaviors and to verify
changes associated with wearing the assistive exoskeleton.
Confidence intervals were presented for all secondary outcomes
as (lower bound, upper bound).
III. RESULTS
A. Residual Hip Effort
The net residual hip energy showed a statistically significant
reduction while wearing the exoskeleton. Specifically, the net
residual hip energy without the exoskeleton was 0.05±0.04 J/kg
(mean ± standard error), decreasing to -0.01±0.05 J/kg with the
Figure 3. Prosthesis side hip torques during the exoskeleton trial, averaged
across all strides and subjects, with shading representing standard error of
means. The left y-axis shows the total residual hip torque (including
exoskeleton torque) (dashed black line) and the biological residual hip
torque (no exoskeleton torque) (dashed red line). The right y-axis shows
the assistive torque applied by the exoskeleton (solid black line).
0 50 100
Phase [%]
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
Hip Torque [Nm/kg]
-0.1
0
0.1
0.2
0.3
0.4
0.5
Assistive Exo Torque [Nm/kg]
Exo
Hip
Figure 4. Secondary analysis of balance and stability. (a) Center of mass in the medial-lateral direction and the anterior-posterior direction, averaged across all
strides and subjects. (b) Center of pressure averaged across all strides and subjects, plotted against the extrapolated center of mass position at heel strike
averaged across all strides and subjects. Error bars represent standard error of means. (c) Minimum distance from the center of mass to the extrapolated center
of mass for each side and condition. (d) Average distance between the sound side and prosthesis side center of pressure (COP) between exoskeleton and no
exoskeleton conditions. (e) Total medial-lateral range of motion while wearing the exoskeleton compared to no exoskeleton. (c)-(e) Display the means across
all strides and subjects (bar heights) and individual subject means (circles).
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
6
exoskeleton (p = 0.026, Figure 2 (a)). The observed difference
in net residual hip energy resulted from visible changes in both
concentric and eccentric energy during different phases of gait.
The 95% confidence intervals for all energy and torque are
reported in Table II; percent differences are discussed below.
On average, with the exoskeleton, concentric energy in the H1
phase decreased by 35.9%, eccentric energy in the H2 phase
increased by 14.0%, and concentric energy in the H3 phase
decreased by 14.8% (Figure 2 (b), Table II). Also, with the
exoskeleton, we observed a 37.5% decrease in biological
residual hip extension torque occurring immediately after heel
strike, and the peak biological flexion torque decreased by 2.1%
(Figure 2 (c), Table II). Similarly, concentric biological power
in the residual hip decreased immediately after heel strike (H1
phase) by 35.0% on average (Figure 2(d), Table II). Biological
peak eccentric power of the residual hip (H2 phase) increased
by 13.9% (Figure 2(d), Table II). Finally, biological peak
concentric power of the residual hip (H3 phase) decreased
slightly by 3.0% (Figure 2(d), Table II).
On average, the exoskeleton provided 0.10±0.01 Nm/kg of
flexion assistance, peaking at 62.4% stride. Also, on average,
the exoskeleton provided 0.06±0.01 Nm/kg of extension
assistance, peaking at 13.9% stride (Figure 3). Figure 3 shows
the mean total residual hip torque profile calculated using
inverse dynamics (black dashed line) superimposed on the
mean assistive torque profile provided by the exoskeleton
(black, solid line) and the mean biological torque profile of the
residual hip (red, dashed line).
B. Center of Mass, Center of Pressure, and Kinematics
The trajectory of the medio-lateral and antero-posterior
center of mass (COM) was not substantially affected by the
exoskeleton assistance (Figure 4(a)). Only small differences
were observed in the lateral center of pressure (COP) between
the exoskeleton condition (1.0 cm more lateral, prosthesis side;
0.2 cm more lateral, sound side) compared to the no
exoskeleton condition (Figure 4 (b)). Similarly, the extrapolated
COM at heel strike was 0.34 cm more lateral on the prosthesis
side and 0.78 cm more lateral on the sound side when using the
exoskeleton compared to without it (Figure 4 (b)). Figure 4 also
shows the minimum distance between extrapolated COM and
average COP (c), the average distance between left and right
Figure 5. Ankle, knee, and hip kinematics for the prosthesis side (top row) and sound side (bottom row) during the exoskeleton (dashed red line) and no
exoskeleton (solid grey line) conditions, averaged across all strides and subjects. Shaded areas represent the standard error of means.
0 50 100
Stride [%]
0
10
20
Position [deg]
Ankle
0 50 100
Stride [%]
-60
-40
-20
0
Prosthesis Side
Knee
0 50 100
Stride [%]
-20
0
20
40
Hip
No Exo Exo
0 50 100
Stride [%]
-20
-10
0
Position [deg]
Ankle
0 50 100
Stride [%]
-60
-40
-20
0
Sound Side
Knee
0 50 100
Stride [%]
-20
0
20
40
Hip
Table III.
Confidence intervals of center of mass (COM), center of pressure
(COP), range of motion (ROM), and joint velocity for sound side and
prosthesis side
Prosthesis Side Sound Side
Peak Medio-
Lateral COM [cm]
Exo (6.3, 6.9)
No Exo (6.1, 6.6)
Minimum
Distance from
Extrapolated COM
to COP [cm]
Exo (6.9, 7.7) (7.7, 8.3)
No Exo (6.5, 7.5) (7.4, 8.2)
Ankle ROM [°]
Exo (14.3, 22.9) (26.9, 33.5)
No Exo (14.5 22.4) (26.5, 31.3)
Knee ROM [°]
Exo (56.4 70.9) (61.7, 69.2)
No Exo (53.9, 68.3) (61.7, 69.0)
Hip ROM [°]
Exo (42.8, 50.8) (37.0, 46.5)
No Exo (42.7, 49.0) (38.4, 47.2)
Peak Flexion Knee
Velocity [°/s]
Exo (385.7, 410.3) (399.6, 465.0)
No Exo (363.3, 386.7) (395.5, 463.3)
Peak Extension
Hip Velocity [°/s]
Exo (196.2, 207.8) (177.2, 188.8)
No Exo (183.7, 196.3) (185.6, 198.4)
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
7
COP (d), and the average medio-lateral center of mass range
(e). Only small differences, on the order of 4%, were observed
in COM- or COP-related measures between tested conditions.
Table III shows 95% confidence intervals for COM and COP
with and without the exoskeleton.
The kinematic profiles for both the prosthesis side and the
sound side, with and without the exoskeleton, are shown in
Figure 5. We reported the 95% confidence intervals for all
kinematic variables in Table III. Similar ranges of motion
(ROM) for all joints were observed with and without the
exoskeleton, and differences between conditions were lower
than 4.5%. Peak extension velocity of the residual hip increased
by 6.3% with the exoskeleton. Moreover, peak extension
velocity of the sound side hip decreased by 4.7% with the
exoskeleton. Peak flexion velocity of the prosthesis side knee
increased by 6.1% with the exoskeleton. Peak flexion velocity
of the sound side knee was minimally affected by the
exoskeleton. Peak flexion and extension velocities of both
prosthesis and sound ankles were not visibly affected by the
exoskeleton.
C. Spatiotemporal Analysis
Spatiotemporal parameters for both the prosthesis side and
sound side were mostly unaffected by the exoskeleton
assistance. Observed differences in spatiotemporal parameters
were lower than 3%, as seen in Table IV, where we report the
95% confidence intervals for all spatiotemporal parameters for
both the prosthesis side and sound side, with and without the
exoskeleton.
IV. DISCUSSION
Powered hip exoskeletons have shown the ability to improve
meaningful clinical outcomes for individuals with above-knee
amputations, including walking economy [33], stumble
recovery [31], and self-selected walking speed after
exoskeleton mediated training [30]. However, previous studies
have not provided a mechanistic explanation for how the
assistance provided by a powered hip exoskeleton affects
amputee gait biomechanics during walking. This study tested
the hypothesis that the assistance provided by the hip
exoskeleton decreases the torque generated by the residual limb
muscles of the hip joint, reducing the residual limb effort. Our
experiments with eight subjects with above-knee amputations
supported this hypothesis by showing that the net residual hip
energy, calculated as the integral of the biological hip power
over the stride duration, significantly decreased by 80% when
walking with the exoskeleton, compared to walking without the
exoskeleton. The analysis of the hip energy during different gait
phases showed that the powered hip exoskeleton had a more
noticeable effect on the concentric residual hip energy than on
the eccentric residual hip energy. This analysis suggests that the
observed reduction in net residual hip energy is mainly due to
reduction in the concentric energy generated by the residual
limb muscles.
This study suggests that the reduction in net residual hip
energy is a major factor in explaining the walking economy
improvements previously observed [33]. We believe that the
reduction in net residual hip energy offers a plausible
mechanism to explain observed metabolic improvements,
because the hip joint has been shown to have low efficiency in
converting metabolic to mechanical energy in healthy
individuals [47]. We also expect that the residual hip joint in
above-knee amputee individuals has a much lower efficiency
than in nonamputees, given the substantial shortening and non-
physiological arrangement of the residual hip muscles after
amputation [10]. Therefore, reducing the mechanical effort at
the hip could have a substantial impact on metabolic energy
during walking.
The analysis of subject-specific results provided additional
insights into how the hip mechanics changed with the
exoskeleton assistance. While the average net residual hip
energy decreased significantly, achieving close to zero sample
mean, for some subjects the magnitude of net residual hip
energy increased – for these subjects, the net residual hip energy
was negative without the exoskeleton, and became more
negative with the exoskeleton. Secondary analysis showed that,
for these subjects, the net residual hip energy became more
negative because of a concurrent reduction of the positive
concentric hip energy (H1 and H3) and increase (in magnitude)
of the negative eccentric hip energy (H2). We believe that this
result agrees with the metabolic cost reduction observed in our
previous study because positive energy generation has higher
metabolic cost than negative energy absorption at the muscle
level [48].
The powered hip exoskeleton reduced the peak hip extension
torque of the residual limb during early stance by 37.5% (Figure
2 (c)). The reduction in hip extension torque could be
particularly helpful because the residual hip extensors in
individuals with above-knee amputations are generally weak
[11], [49]. Interestingly, there were no visible differences in
residual hip peak flexion torque despite the residual hip flexors
receiving substantial assistance from the exoskeleton. This
result may be due to the timing of the hip flexion assistance,
which seems to be slightly late compared to the torque produced
by the residual limb (Figure 3). Thus, although the powered hip
exoskeleton assisted the residual limb both in flexion and
extension, the extension assistance seems to be the main
contributor to the observed reduction in residual limb energy
and effort. Further tests are necessary to assess the effect of
timing and magnitude of assistance in flexion and extension on
biomechanical adaptations.
In this study, we chose to assist only the residual limb using
an asymmetric, unilateral exoskeleton configuration. Choosing
to assist the residual limb might have further exacerbated
existing kinematic and kinetic compensatory strategies or
affected COM and COP trajectories. However, there were no
meaningful changes to the antero-posterior or medio-lateral
motion of the COM when participants used the exoskeleton
(Figure 4). Compared to nonamputees, above-knee amputees
walk with a wider center of pressure between limbs [3]. We
observed no meaningful changes in the COP of either limb
relative to the extrapolated COM while walking (Figure 4 (b)).
Thus, the analysis of the center of mass (COM) and center of
pressure (COP) suggests that the exoskeleton assistance does
not have any apparent effects on balance. The kinematic
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
8
analysis shows only minor differences between walking with
and without the exoskeleton. Also, there were no visible
changes in ROM on either limb, which does not support our
previous study in which a significant increase in the ROM of
the residual limb was reported [33]. Although no changes in
joint angles were observed, we did identify significant changes
in velocity. The residual hip extension velocity increased by
6.3% and the sound hip extension velocity decreased by 4.7%
when wearing the exoskeleton. Although residual hip extension
velocity increased, no corresponding increase in concentric
extension power was observed. In fact, peak concentric residual
hip extension power decreased due to the decrease in extension
torque. This secondary kinematic analysis provides interesting
insights into the adaptation mechanisms observed during the
powered hip exoskeleton assistance. However, the observed
changes are generally small and, therefore, do not seem to play
a major role in explaining how above-knee amputees adapt to
powered hip exoskeletons.
The effects of the powered hip exoskeleton on spatiotemporal
parameters and gait symmetry were minimal and unlikely to be
clinically meaningful. The study participants showed relatively
symmetric gait (Table IV, Figure 4, Figure 5). This result is not
surprising given that the inclusion criteria required subjects to
be able to walk on a treadmill without handrails, which is only
possible for above-knee amputees with a high level of mobility.
On the other hand, based on the observed changes in joint
velocities, we would have expected to see more substantial
changes in stance time, swing time, or step lengths.
A. Limitations
A limitation of this study is that we were not able to include
electromyography. Understanding the influence of the
exoskeleton on muscle activations would have added important
insights into the mechanisms used to adapt to the exoskeleton
assistance. Notably, we did not observe any meaningful
changes in COM kinematics, which would likely be linked to
changes in lower-back muscle activity. Also, we observed a
reduction in residual limb extension torque, which could
indicate reduced effort in the extensor muscles in the residual
limb. Future work should explore the changes in lower-limb
muscular effort while wearing the hip exoskeleton, similar to
what was recently done in nonamputee individuals [50].
Another limitation of this study is that we assumed a perfect
transmission of torque from the exoskeleton to the user’s
residual hip. Although this is a common assumption made in
the field, it can have an impact on the interpretation of the
results. In a previous study [41], we used a 6-axis loadcell at the
interface between a powered knee exoskeleton and the user to
directly measure how the torque was transmitted from the
exoskeleton to the user. The result of this study suggested that
the assumption of perfect transfer of torque is reasonable
provided that a self-aligning mechanism is implemented in the
exoskeleton. Self-aligning mechanisms can significantly reduce
spurious forces and torques between the user and the
exoskeleton interface, therefore minimizing inefficiencies [41].
They also have a significant effect on performance and comfort
[42]. Because the powered hip exoskeleton used in this study
uses a self-aligning mechanism [51], the assumption of perfect
transmission of forces and torques to the user should not cause
substantial inaccuracies.
Participants with lower mobility or higher asymmetry may
respond to the residual hip assistance provided by the powered
hip exoskeleton in a different way compared to what we
observed in this study. In the present study, participants were
required to walk on a treadmill continuously without using the
handrails for assistance. Consequently, the participants were of
a high mobility index, which limits our understanding of how
the exoskeleton might change kinematics and kinetics of
individuals with a low mobility index. Future work should
assess the impact of unilateral hip assistance in a lower mobility
population, because the hip exoskeleton might prove even more
beneficial to lower mobility populations.
Another limitation of this study is that we did not use any
subjective measures of effort, such as rated perceived exertion.
Our study participants subjectively commented on the
assistance from the hip exoskeleton, likening it to using an
electric bicycle. In fact, many participants commented on the
difficulty of walking after doffing the exoskeleton. These
subjective remarks are encouraging, but do not offer qualitative
information regarding exoskeleton assistance outside of
laboratory environments. Future work should include
subjective analyses regarding user preferences and real-world
applications.
Finally, understanding how assistive devices interact with
prescribed passive prostheses is an important step in making
powered exoskeletons a clinically viable intervention.
Changing user prosthesis alignment might influence the
interaction between the prosthesis and exoskeleton. Prosthesis
tuning was not explored as a variable in this study and
represents a limitation.
Table IV.
Confidence intervals of spatiotemporal parameters
Prosthesis Side Sound Side
Stride Time [s]
Exo (1.16, 1.18) (1.16, 1.18)
No Exo (1.16, 1.18) (1.16, 1.18)
Stance Time [s]
Exo (0.72, 0.74) (0.79, 0.80)
No Exo (0.70, 0.72) (0.79, 0.80)
Swing Time [s]
Exo (0.44, 0.46) (0.36, 0.38)
No Exo (0.45, 0.47) (0.36, 0.38)
Cadence
[steps/min]
Exo (94.8, 97.0) (108.3, 111.7)
No Exo (94.9, 97.3) (108.5, 112.1)
Step Length [m]
Exo (0.74, 0.76) (0.55, 0.61)
No Exo (0.74, 0.76) (0.60, 0.62)
Step Width [m]
Exo (0.23, 0.24)
No Exo (0.23, 0.24)
Minimum
Prosthesis Side
Toe Clearance
[cm]
Exo (7.00, 7.66)
No Exo (6.95, 7.35)
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
9
V. CONCLUSION
In this paper, we evaluated the effect of a unilateral,
lightweight powered hip exoskeleton on the gait biomechanics
in eight above-knee amputees. The powered exoskeleton
assistance significantly reduced the net residual hip energy,
with the concentric H1 energy (0-30% gait cycle) being the
main contributor to this change. Moreover, the powered
exoskeleton assistance visibly affected the residual hip
extension torque. Notably, the powered exoskeleton assistance
had negligible effects on joint angles, base of support, and
centers of pressure. This result indicates that the powered
exoskeleton did not negatively affect balance and stability.
Future work should aim to improve balance and stability, which
are typically poor in individuals with above-knee amputations.
We believe that this outcome could be achieved by assisting the
residual hip joint in abduction and adduction to provide frontal
plane stability and assistance. Future work should also aim to
assess long-term efficacy in a take-home trial or understand the
importance of an assistive profile based on the needs of the
population.
REFERENCES
[1] A. D. Segal et al., “Kinematic and kinetic comparisons
of transfemoral amputee gait using C-Leg and Mauch
SNS prosthetic knees,” The Journal of Rehabilitation
Research and Development, vol. 43, no. 7, p. 857, 2006,
doi: 10.1682/JRRD.2005.09.0147.
[2] Y. Sagawa, K. Turcot, S. Armand, A. Thevenon, N.
Vuillerme, and E. Watelain, “Biomechanics and
physiological parameters during gait in lower-limb
amputees: A systematic review,” Gait Posture, vol. 33,
no. 4, pp. 511–526, Apr. 2011, doi:
10.1016/j.gaitpost.2011.02.003.
[3] A. L. Hof, R. M. van Bockel, T. Schoppen, and K.
Postema, “Control of lateral balance in walking,” Gait
Posture, vol. 25, no. 2, pp. 250–258, Feb. 2007, doi:
10.1016/j.gaitpost.2006.04.013.
[4] T. Schmalz, S. Blumentritt, and R. Jarasch, “Energy
expenditure and biomechanical characteristics of lower
limb amputee gait:,” Gait Posture, vol. 16, no. 3, pp.
255–263, Dec. 2002, doi: 10.1016/S0966-
6362(02)00008-5.
[5] R. Gailey, K. Allen, J. Castles, J. Kucharik, and M.
Roeder, “Review of secondary physical conditions
associated with lower-limb amputation and long-term
prosthesis use,” The Journal of Rehabilitation Research
and Development, vol. 45, no. 1, pp. 15–30, Dec. 2008,
doi: 10.1682/JRRD.2006.11.0147.
[6] D. M. Ehde, D. G. Smith, J. M. Czerniecki, K. M.
Campbell, D. M. Malchow, and L. R. Robinson, “Back
pain as a secondary disability in persons with lower limb
amputations,” Arch Phys Med Rehabil, vol. 82, no. 6, pp.
731–734, Jun. 2001, doi: 10.1053/apmr.2001.21962.
[7] K. Sasaki, R. R. Neptune, and S. A. Kautz, “The
relationships between muscle, external, internal and
joint mechanical work during normal walking.,” J Exp
Biol, vol. 212, no. Pt 5, pp. 738–744, 2009, doi:
10.1242/jeb.023267.
[8] D. J. Farris and G. S. Sawicki, “The mechanics and
energetics of human walking and running: a joint level
perspective,” Journal of The Royal Society Interface,
vol. 9, no. 66. pp. 110–118, 2012. doi:
10.1098/rsif.2011.0182.
[9] T. S. Bae, K. Choi, D. Hong, and M. Mun, “Dynamic
analysis of above-knee amputee gait,” Clinical
Biomechanics, vol. 22, no. 5, pp. 557–566, Jun. 2007,
doi: 10.1016/j.clinbiomech.2006.12.009.
[10] F. a Gottschalk and M. Stills, “The biomechanics of
trans-femoral amputation.,” Prosthet Orthot Int, vol. 18,
no. 1, pp. 12–17, 1994, doi:
10.3109/03093649409164665.
[11] R. R. E. Seroussi, A. Gitter, J. M. Czerniecki, and K.
Weaver, “Mechanical work adaptations of above-knee
amputee ambulation,” Arch Phys Med Rehabil, vol. 77,
no. 11, pp. 1209–1214, Nov. 1996, doi: 10.1016/S0003-
9993(96)90151-3.
[12] E. H. Sinitski, A. H. Hansen, and J. M. Wilken,
“Biomechanics of the ankle–foot system during stair
ambulation: Implications for design of advanced ankle–
foot prostheses,” J Biomech, vol. 45, no. 3, pp. 588–594,
Feb. 2012, doi: 10.1016/j.jbiomech.2011.11.007.
[13] M. Rabuffetti, M. Recalcati, and M. Ferrarin, “Trans-
Femoral Amputee Gait,” Prosthet Orthot Int, vol. 29, no.
2, pp. 183–192, Aug. 2005, doi:
10.1080/03093640500217182.
[14] H. Goujon-Pillet, E. Sapin, P. Fodé, and F. Lavaste,
“Three-Dimensional Motions of Trunk and Pelvis
During Transfemoral Amputee Gait,” Arch Phys Med
Rehabil, vol. 89, no. 1, pp. 87–94, Jan. 2008, doi:
10.1016/j.apmr.2007.08.136.
[15] S. Hood, M. K. Ishmael, A. Gunnell, K. B. Foreman, and
tommaso Lenzi, “A kinematic and kinetic dataset of 18
above-knee amputees walking at various speeds,” Sci
Data, vol. 7, no. 1, pp. 1–8, 2020, doi: 10.1038/s41597-
020-0494-7.
[16] E. Russell Esposito and J. M. Wilken, “The relationship
between pelvis–trunk coordination and low back pain in
individuals with transfemoral amputations,” Gait
Posture, vol. 40, no. 4, pp. 640–646, Sep. 2014, doi:
10.1016/j.gaitpost.2014.07.019.
[17] J. L. Johansson, D. M. Sherrill, P. O. Riley, P. Bonato,
and H. Herr, “A Clinical Comparison of Variable-
Damping and Mechanically Passive Prosthetic Knee
Devices,” Am J Phys Med Rehabil, vol. 84, no. 8, pp.
563–575, Aug. 2005, doi:
10.1097/01.phm.0000174665.74933.0b.
[18] K. R. Kaufman et al., “Gait and balance of transfemoral
amputees using passive mechanical and microprocessor-
controlled prosthetic knees,” Gait Posture, vol. 26, no.
4, pp. 489–493, Oct. 2007, doi:
10.1016/j.gaitpost.2007.07.011.
[19] L. L. McNealy and S. A. Gard, “Effect of prosthetic
ankle units on the gait of persons with bilateral trans-
femoral amputations,” Prosthet Orthot Int, vol. 32, no.
1, pp. 111–126, Mar. 2008, doi:
10.1080/02699200701847244.
[20] M. L. van der Linden, S. E. Solomonidis, W. D. Spence,
N. Li, and J. P. Paul, “A methodology for studying the
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
10
effects of various types of prosthetic feet on the
biomechanics of trans-femoral amputee gait,” J
Biomech, vol. 32, no. 9, pp. 877–889, Sep. 1999, doi:
10.1016/S0021-9290(99)00086-X.
[21] B. E. Lawson, H. A. Varol, A. Huff, E. Erdemir, and M.
Goldfarb, “Control of Stair Ascent and Descent With a
Powered Transfemoral Prosthesis,” IEEE Transactions
on Neural Systems and Rehabilitation Engineering, vol.
21, no. 3, pp. 466–473, May 2013, doi:
10.1109/TNSRE.2012.2225640.
[22] S. A. Hood and T. Lenzi, “Preliminary Analysis Of
Positive Knee Energy Injection In A Transfemoral
Amputee Walking With A Powered Prosthesis,” in 2018
40th Annual International Conference of the IEEE
Engineering in Medicine and Biology Society (EMBC),
2018, vol. 2018-July, pp. 1821–1824. doi:
10.1109/EMBC.2018.8512726.
[23] R. C. Browning, J. R. Modica, R. Kram, and A.
Goswami, “The effects of adding mass to the legs on the
energetics and biomechanics of walking,” Med Sci
Sports Exerc, vol. 39, no. 3, pp. 515–525, 2007, doi:
10.1249/mss.0b013e31802b3562.
[24] R. W. Selles, J. B. J. Bussmann, R. C. Wagenaar, and H.
J. Stam, “Effects of prosthetic mass and mass
distribution on kinematics and energetics of prosthetic
gait: A systematic review,” Arch Phys Med Rehabil, vol.
80, no. 12, pp. 1593–1599, 1999, doi: 10.1016/S0003-
9993(99)90336-2.
[25] J. D. Smith and P. E. Martin, “Effects of Prosthetic Mass
Distribution on Metabolic Costs and Walking
Symmetry,” J Appl Biomech, vol. 29, no. 3, pp. 317–328,
Jun. 2013, doi: 10.1123/jab.29.3.317.
[26] J. D. Smith and P. E. Martin, “Short and Longer Term
Changes in Amputee Walking Patterns Due to Increased
Prosthesis Inertia,” JPO Journal of Prosthetics and
Orthotics, vol. 23, no. 3, pp. 114–123, Jul. 2011, doi:
10.1097/JPO.0b013e3182248d90.
[27] M. Tran, L. Gabert, M. Cempini, and T. Lenzi, “A
Lightweight, Efficient Fully Powered Knee Prosthesis
With Actively Variable Transmission,” IEEE Robot
Autom Lett, vol. 4, no. 2, pp. 1186–1193, Apr. 2019, doi:
10.1109/LRA.2019.2892204.
[28] L. Gabert, S. Hood, M. Tran, M. Cempini, and T. Lenzi,
“A Compact, Lightweight Robotic Ankle-Foot
Prosthesis: Featuring a Powered Polycentric Design,”
IEEE Robot Autom Mag, vol. 27, no. 1, pp. 87–102, Mar.
2020, doi: 10.1109/MRA.2019.2955740.
[29] E. Martini et al., “Lower-limb amputees can reduce the
energy cost of walking when assisted by an Active Pelvis
Orthosis,” in Proceedings of the IEEE RAS and EMBS
International Conference on Biomedical Robotics and
Biomechatronics, 2020, vol. 2020-Novem. doi:
10.1109/BioRob49111.2020.9224417.
[30] C. B. Sanz-Morère et al., “Robot-mediated overground
gait training for transfemoral amputees with a powered
bilateral hip orthosis: a pilot study,” J Neuroeng Rehabil,
vol. 18, no. 1, p. 111, Dec. 2021, doi: 10.1186/s12984-
021-00902-7.
[31] M. Vito et al., “Counteracting Balance Loss in
Transfemoral Amputees by Using an Active Pelvis
Orthosis: A Case Series,” 2021, pp. 294–305. doi:
10.1007/978-3-030-64610-3_35.
[32] M. K. Ishmael, M. Tran, and T. Lenzi, “ExoProsthetics:
Assisting Above-Knee Amputees with a Lightweight
Powered Hip Exoskeleton,” in 2019 IEEE 16th
International Conference on Rehabilitation Robotics
(ICORR), Jun. 2019, vol. 2019-June, pp. 925–930. doi:
10.1109/ICORR.2019.8779412.
[33] M. K. Ishmael, D. Archangeli, and T. Lenzi, “Powered
hip exoskeleton improves walking economy in
individuals with above-knee amputation,” Nat Med, vol.
27, no. 10, pp. 1783–1788, Oct. 2021, doi:
10.1038/s41591-021-01515-2.
[34] R. M. Ehrig, W. R. Taylor, G. N. Duda, and M. O.
Heller, “A survey of formal methods for determining the
centre of rotation of ball joints,” J Biomech, vol. 39, no.
15, pp. 2798–2809, Jan. 2006, doi:
10.1016/j.jbiomech.2005.10.002.
[35] W. Taylor, R. Ehrig, G. Duda, and M. Heller, “The
symmetrical axis of rotation approach (SARA) for
determination of joint axes in clinical gait analysis,” Gait
Posture, vol. 24, pp. S18–S20, Dec. 2006, doi:
10.1016/j.gaitpost.2006.11.017.
[36] S. v. Sarkisian, M. K. Ishmael, G. R. Hunt, and T. Lenzi,
“Design, Development, and Validation of a Self-
Aligning Mechanism for High-Torque Powered Knee
Exoskeletons,” IEEE Trans Med Robot Bionics, vol. 2,
no. 2, pp. 248–259, May 2020, doi:
10.1109/TMRB.2020.2981951.
[37] S. v. Sarkisian, M. K. Ishmael, and T. Lenzi, “Self-
Aligning Mechanism Improves Comfort and
Performance With a Powered Knee Exoskeleton,” IEEE
Transactions on Neural Systems and Rehabilitation
Engineering, vol. 29, pp. 629–640, 2021, doi:
10.1109/TNSRE.2021.3064463.
[38] T. Lenzi, M. C. Carrozza, and S. K. Agrawal, “Powered
hip exoskeletons can reduce the user’s hip and ankle
muscle activations during walking,” IEEE Transactions
on Neural Systems and Rehabilitation Engineering, vol.
21, no. 6, pp. 938–948, 2013, doi:
10.1109/TNSRE.2013.2248749.
[39] R. Ronsse, N. Vitiello, T. Lenzi, J. van den Kieboom, M.
C. Carrozza, and A. J. Ijspeert, “Human–Robot
Synchrony: Flexible Assistance Using Adaptive
Oscillators,” IEEE Trans Biomed Eng, vol. 58, no. 4, pp.
1001–1012, Apr. 2011, doi:
10.1109/TBME.2010.2089629.
[40] R. Ronsse et al., “Oscillator-based assistance of cyclical
movements: Model-based and model-free approaches,”
Med Biol Eng Comput, vol. 49, no. 10, pp. 1173–1185,
Oct. 2011, doi: 10.1007/s11517-011-0816-1.
[41] D. A. Winter, Biomechanics and Motor Control of
Human Gait: Normal, Elderly and Pathological, 2nd ed.
Waterloo, Ontario: Waterloo Biomechanics Press, 1991.
[42] D. J. Farris and G. S. Sawicki, “Linking the mechanics
and energetics of hopping with elastic ankle
exoskeletons,” J Appl Physiol, vol. 113, no. 12, pp.
1862–1872, Dec. 2012, doi:
10.1152/japplphysiol.00802.2012.
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
11
[43] S. H. Collins, M. B. Wiggin, and G. S. Sawicki,
“Reducing the energy cost of human walking using an
unpowered exoskeleton,” Nature, vol. 522, no. 7555, pp.
212–215, Jun. 2015, doi: 10.1038/nature14288.
[44] G. S. Sawicki and D. P. Ferris, “Mechanics and
energetics of level walking with powered ankle
exoskeletons,” Journal of Experimental Biology, vol.
211, no. 9, pp. 1402–1413, May 2008, doi:
10.1242/jeb.009241.
[45] A. L. Hof, “The ‘extrapolated center of mass’ concept
suggests a simple control of balance in walking,” Hum
Mov Sci, vol. 27, no. 1, pp. 112–125, Feb. 2008, doi:
10.1016/j.humov.2007.08.003.
[46] A. L. Hof, M. G. J. Gazendam, and W. E. Sinke, “The
condition for dynamic stability,” J Biomech, vol. 38, no.
1, pp. 1–8, Jan. 2005, doi:
10.1016/j.jbiomech.2004.03.025.
[47] D. P. F. Gregory S. Sawicki, Cara L. Lewis, “It pays to
have spring in your step,” Exerc Sport Sci Rev., vol. 37,
no. 3, pp. 130–137, 2009, doi: 10.1038/jid.2014.371.
[48] D. J. Farris and G. S. Sawicki, “The mechanics and
energetics of human walking and running: a joint level
perspective,” J R Soc Interface, vol. 9, no. 66, pp. 110–
118, Jan. 2012, doi: 10.1098/rsif.2011.0182.
[49] H. Burger, V. Valenčič, Č. Marinček, and N. Kogovšek,
“Properties of musculus gluteus maximus in above-knee
amputees,” Clinical Biomechanics, vol. 11, no. 1, pp.
35–38, Jan. 1996, doi: 10.1016/0268-0033(95)00032-1.
[50] J. A. George et al., “Robust Torque Predictions From
Electromyography Across Multiple Levels of Active
Exoskeleton Assistance Despite Non-linear
Reorganization of Locomotor Output,” Front
Neurorobot, vol. 15, Nov. 2021, doi:
10.3389/fnbot.2021.700823.
[51] M. K. Ishmael, D. Archangeli, and T. Lenzi, “A Powered
Hip Exoskeleton With High Torque Density for
Walking, Running, and Stair Ascent,” IEEE/ASME
Transactions on Mechatronics, pp. 1–12, 2022, doi:
10.1109/TMECH.2022.3159506.
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
12
VI. SUPPLEMENTARY MATERIALS
A. Hip Compensatory Movements Without the Exoskeleton
Individuals with above-knee amputations alter their hip
mechanics both on the prosthesis side and sound side to
compensate for prosthesis inefficiencies during ambulation. To
assess these compensatory strategies in our study participants,
we analyzed the torque, power, and energy of the sound side
and prosthesis side hip without the exoskeleton. As shown in
Supplementary Figure 1, the extension hip torque (i.e., negative
torque) was generally higher for the sound side (blue line) than
the prosthesis side (gray line) (Supplementary Figure 1 (a)).
Moreover, the power during the H1 phase was generally higher
for the sound side than the prosthesis side (Supplementary
Figure 1 (b)). In contrast, the power during H2 and H3 was
substantially higher for the prosthesis side than the sound side
(Supplementary Figure 1 (b)). These results suggest that there
was increased sound side power in the first part of the gait cycle
and increased prosthesis side power in the middle part of the
gait cycle. Interestingly, there was no clear trend for the
concentric H1 energy, whereas the hip energy during both H2
and H3 was substantially higher for the prosthesis side than the
sound side for all subjects (Supplementary Figure 1 (c)).
B. Total Hip Energy and Power With and Without the
Exoskeleton
Depending on how the user adapts to the exoskeleton
assistance, the total hip power (i.e., the sum of the power
generated by the residual hip and the exoskeleton) can increase
or decrease during ambulation. To further assess the effect of
the exoskeleton assistance, we performed an additional analysis
comparing the total hip power on the prosthesis side with and
without the exoskeleton. The results of this analysis are shown
in Supplementary Figure 2. These results show that the total
concentric H1 hip power was slightly lower with the
exoskeleton. In contrast, both the total eccentric H2 energy and
the total concentric H3 energy were visibly higher with the
exoskeleton. Thus, the exoskeleton assistance seems to cause
the total hip H2 and H3 power to increases rather than the
residual limb H2 and H3 power to decrease.
Supplementary Figure 1 Analysis of hip torque, power, and energy without
the exoskeleton. (a) Hip torque profiles averaged across all strides and
subjects (b) Hip power profiles averaged across all strides and subjects (c)
Phase energy displaying the averages across all strides and subjects (bar
heights) and individual subject means (circles).
0 50 100
-1
-0.5
0
0.5
1
1.5
Hip Torque [Nm/kg]
Sound Side No Exo
Prosthesis Side No Exo
H1
H2
H3
0 50 100
Stride [%]
-1.5
-1
-0.5
0
0.5
1
1.5
Hip Power [W/kg]
0
0.05
0.1
0.15
0.2
Concentric
H1 Energy [J/kg]
0
0.05
0.1
0.15
0.2
Eccentric
H2 Energy [J/kg]
Sound Side
No Exo
Prosthesis Side
No Exo
0
0.05
0.1
0.15
0.2
Concentric
H3 Energy [J/kg]
(c)(b)
(a)
Supplementary Figure 2 Analysis of torque, power, and energy of the
prosthesis side hip during the no exoskeleton condition (green), and the
total torque, power and energy of the prosthesis side hip and exoskeleton
combined during the exoskeleton condition (pink). (a) Hip torque profiles
averaged across all strides and subjects (b) Hip power profiles averaged
across all strides and subjects (c) Phase energy, displaying the averages
across all strides and subjects (bar heights) and individual subject data
points (circles).
0 50 100
-0.5
0
0.5
1
1.5
Total Prosthesis Side
Hip Torque [Nm/kg]
No Exo
With Exo
H1
H2
H3
0 50 100
Stride [%]
-1.5
-1
-0.5
0
0.5
1
1.5
Total Prosthesis Side
Hip Power [W/kg]
0
0.1
0.2
Concentric
H1 Energy [J/kg]
0
0.1
0.2
0.3
Eccentric
H2 Energy [J/kg]
No Exo
With Exo
0
0.05
0.1
0.15
0.2
Concentric
H3 Energy [J/kg]
(b) (c)
(a)
Supplementary Figure 3 Torque (top) and power (bottom) plots of the three
subjects using a C-Leg (blue) and three subjects using a Plie (green).
Dashed lines show the no exoskeleton condition while solid lines show the
exoskeleton condition. Data were averaged across all strides and subjects.
0 20 40 60 80 100
-0.5
0
0.5
1
1.5
Prosthesis Side Hip
Torque with Exo [Nm/kg]
Plie With Exo
C-Leg With Exo
Plie No Exo
C-Leg No Exo
0 20 40 60 80 100
Phase [%]
-2
-1
0
1
2
Prosthesis Side Hip
Power with Exo [W/kg]
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
13
C. Effect of Prescribed Prosthesis
A subset of our population used a C-Leg (n = 3 participants),
and a different subset of the used a Plie (n = 3 participants). We
performed an analysis to assess the influence of prescribed
prosthesis on hip torque and power on their prosthesis side hip
(Supplementary Figure 3). Interestingly, study participants
using the Plie had a higher peak power in the H2 and H3 phases
of the gait cycle compared to study participants using the C-
Leg. However, these observed differences might be a result of
their differing demographics. We would need a larger patient
population to control for these variables in order to truly
understand the differences between these groups in terms of
prosthesis side hip exoskeleton assistance. However, neither
subset demonstrated large changes between the exoskeleton and
no exoskeleton conditions in either of these phases. Instead,
both groups display a trend that agrees with the overall trend
observed in the paper – a decreased extension torque and power
in the H1 phase of the gait cycle.
This article has been accepted for publication in IEEE Transactions on Biomedical Engineering. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/TBME.2022.3211842
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
