robotics.sciencemag.org/cgi/content/full/2/2/eaah4416/DC1
Supplementary Materials for
Assistance magnitude versus metabolic cost reductions for a tethered
multiarticular soft exosuit
B. T. Quinlivan, S. Lee, P. Malcolm, D. M. Rossi, M. Grimmer, C. Siviy, N. Karavas,
D. Wagner, A. Asbeck, I. Galiana, C. J. Walsh*
*Corresponding author. Email: [email protected]
Published 18 January 2017, Sci. Robot. 2, eaah4416 (2016)
DOI: 10.1126/scirobotics.aah4416
The PDF file includes:
Supplementary Text
Fig. S1. Structure of the waist belt component.
Fig. S2. Structure of the calf wrap component.
Fig. S3. Structure of the Y-strap part of the calf wrap component.
Fig. S4. Structure of the vertical strap component.
Fig. S5. Changes in ground reaction forces and center of mass velocity.
Fig. S6. Changes in center of mass power.
Fig. S7. Hip and ankle force profiles.
Fig. S8. Changes in biological joint moment.
Fig. S9. Changes in biological joint power.
Fig. S10. Net biological joint power.
Table S1. Changes in net metabolic rate for each participant.
References (42–44)
Other Supplementary Material for this manuscript includes the following:
(available at robotics.sciencemag.org/cgi/content/full/2/2/eaah4416/DC1)
Movie S1 (.mp4 format). Varying assistance level with a soft exosuit: Methods
and metabolic results.
Soft exosuit design
Several design principles were used when developing these textile components as previously highlighted in
Asbeck et al. (37). In brief, the suits were designed to minimize component drift, maximize the power transfer
to the human, and maximize user comfort. To achieve these goals the following design principles were used: (i)
leveraging human geometry to maximize stiffness and minimize component drift, (ii) anchoring the suit to
locations of the body with high stiffness, (iii) sourcing materials with high stiffness but good conformability,
(iv) routing forces across the body using the most direct path, (v) minimizing shear forces between the suit and
human to improve comfort, and (vi) maximizing the suit-skin contact area to minimize normal pressure and
maximize friction between the suit and human skin.
As noted in Materials and Methods, the textile components of the exosuit used in this study consisted of a
spandex base layer, a waist belt, two calf wraps, and four vertical straps (two per leg) crossing from the back of
the calf wrap, through the center of the knee joint axis, to the front of the waist belt. An overview of how all
these components fit into the full system is shown in Fig 1. The base layer was a simple spandex designed to
maximize breathability. Detailed images of each of the other textile components (waist belt, calf wrap and the
accompanying Y-strap, and vertical straps) along with the materials used in their construction are shown in figs.
S1 to S4. These components attach to a load cell (LTH300, FUTEK Advanced Sensor Technology) and
Bowden cable via a custom built mechanical assembly and two quick release pins which thread through (i) the
bottom of the calf wrap and (ii) the bottom of the Y-strap of the calf wrap and the polyester cord which
connects the Y-strap to the vertical straps.
Center of mass power analysis
Center of mass power was calculated for the right limb as described by Donelan et al. (42). Average push-off
power increased with increasing exosuit assistance while average collision and rebound power decreased and
average preload power remained unchanged (fig. S6).
With a unilateral exoskeleton Jackson and Collins (23) found that with increasing exoskeleton work, average
push-off and average collision power increased for the assisted limb and average contralateral push-off,
collision, and rebound power decreased. Jackson and Collins also found that with the same unilateral system,
increasing exoskeleton torque resulted in the several opposite trends in that average push-off power decreased
for the assisted limb and average contralateral collision and rebound power increased. Thus, our results
somewhat align with those of Jackson and Collins, but it is difficult to compare due to the unilateral nature of
their system in addition to the fact that both exosuit torque and work increased in our sweep study.
Estimated metabolic cost of wearing the suit components
As the main focus of this lab-based study was understanding the relative effect of assistance level, we chose to
compare suit powered versus unpowered. This simplified the study procedures as it removed the time it takes to
don/doff the suit and the need to remove and reattach motion capture markers. Furthermore, comparing an
unpowered and powered condition eliminated the need to remove and reattach motion capture markers which
could introduce potential sources of error in the analysis comparing kinematics and kinetics across conditions.
However, the question still remains how wearing the suit powered compared to walking without the added
weight of the suit components (textiles, sensors, and attachments) would impact metabolic cost.
For the comparison of the energy cost of walking without the exosuit and in the powered-off condition, we have
estimated the increase in metabolic cost by the weight of the textile, sensors, and attachment components
compared to walking without the exosuit. Using data by Browning et al. (43) we can approximate the metabolic
cost due to the added weight on each segment of the lower extremities (∆𝑀𝐶) to be 0.120 W kg-1. Compared to
the average net metabolic cost of powered-off walking in this study (𝑀𝐶𝑠𝑢𝑖𝑡 ; 4.806 W kg-1), this gives a 2.56%
increase.
∆𝑀𝐶
∆𝑀𝐶
0.120
=
=
= 2.56 %
𝑀𝐶𝑛𝑜 𝑠𝑢𝑖𝑡 𝑀𝐶𝑠𝑢𝑖𝑡 − ∆𝑀𝐶
4.806 − 0.120
This small estimated increase aligns with results from a previous experiment with a similar exosuit but with
different components assisting hip extension during loaded walking. A small increase in metabolic rate (0.10 W
kg-1) was found by wearing those exosuit components, and there was no significant difference between the
powered-off and no-suit conditions [Ding et al. (44)]. For the multiarticular exosuit in this paper, it would be
expected that the measured cost would be higher than found experimentally with the hip-only system in Ding et
al. due to the increased weight (1.163 kg versus 0.683 kg) and different weight distribution (waist, 0.443 kg
versus 0.435 kg; thigh, 0 kg versus 0.248 kg; shank, 0.356 kg versus 0 kg; foot, 0.364 kg versus 0 kg). Using the
same data by Browning et al. along with the weight distribution of the hip extension suit used in Ding et al., the
estimated increase in metabolic rate would be 0.0382 W kg-1. As such we acknowledge that the 2.56% increase
calculated above is an underestimate of the cost but we can assume that the value is between 2.5 to 6.5%.
Fig. S1. Structure of the waist belt component. The waist belt is mainly composed of a plain woven (Typhoon
Black, Source: Springfield) sourced to minimize strain and two different types of Sailcloth Injection Fiber
(Source: Dimension Polyant) were used for reinforcement. Hook and Loop Velcro were used for the front
closure. A custom laminated padding (Source: Dela Foam) was used to line the inside of the belt near the iliac
crests and a small window in the woven material allows for a more conformal fit and reduced pressure
concentration on the anterior superior iliac spine (ASIS). Additionally, four 2” 2-bar slides (Source: YKK) were
sewn into the front of the belt as attachment points for the vertical straps. This waist belt was also designed to
assist with hip extension which is why attachment points and reinforced load paths can be seen on the back of
the waist belt. The 4” Elastic strips (Source: North East Knitting) coming down off the belt are designed to
attach to a thigh brace used when assisting hip extension but no thigh brace or hip extension assistance was
used for this protocol and as a result the elastic straps were simply pinned up onto the belt.
Fig. S2. Structure of the calf wrap component. Similar to the waist belt, the calf wrap is mainly composed of a
plain woven (Typhoon Black, Source: Springfield) and two different types of Sailcloth Injection Fiber (Source:
Dimension Polyant). Additionally, two 1” 2-Bar Slides (Source: YKK), Velcro Hook and Velcro Loop are used
for the top closure. Cord Lacing (Source: Rhode Island Textile), 3/9” Webbing (Source), and Spring Lock
(Source: Strapworks) are used for the lower closer. An insert of custom laminated Padding (Source: Dela
Foam) was used to improve comfort and pressure distribution on the shin. A simple Y-strap (fig. S3) is also part
of the calf wrap and attached back of the main calf wrap via two more 1” 2-Bar Slides (Source: YKK) and used
to connect the wrap to the vertical straps.
Fig S3. Structure of the Y-strap part of the calf wrap component. The Y-strap is a separate portion of the calf
wrap also mainly composed of a plain woven (Typhoon Black, Source: Springfield) and two different types of
Sailcloth Injection Fiber (Source: Dimension Polyant). Velcro Hook and Loop allow for the Y-strap to be
threaded through the 2-Bar Slides on the back of the calf wrap and secured. A polyester cord (Source: R & W
Rope) was used to connect the Y-strap to the vertical straps.
Fig. S4. Structure of the vertical strap component. Four vertical straps are used for each suit, two per leg, and
the straps connect the front of the waist belt to the Y-strap. These simple components are also composed of
mainly composed of a plain woven (Typhoon Black, Source: Springfield) and two different types of Sailcloth
Injection Fiber (Source: Dimension Polyant). Hook and Loop Velcro line the inside of the strap to allow for
high adjustment and a high tensile strength polyester cord is threaded through a plastic D-ring to attach the
strap to Y-strap.
Fig. S5. Changes in ground reaction forces and center of mass velocity. Average (A) medial-lateral, (B)
anterior-posterior, and (C) vertical ground reaction forces over the gait cycle averaged across all participants
for each experimental condition. Average (D) medial-lateral, (E) anterior-posterior, and (F) vertical center of
mass velocity over the gait cycle averaged across all participants for each experimental condition. These
parameters were used to calculate the center of mass power. Gray shading outlines approximate region during
which assistance was applied.
Fig. S6. Changes in center of mass power. (A) The total center of mass power for the right limb normalized to
body mass over the gait cycle for each condition. Center of mass power was defined as the dot product of
ground reaction force with center of mass velocity as described in Donelan et al. (42). Gray shading outlines
approximate region during which assistance was applied. Bar graphs on the right indicates the average power
during certain portions of the gait cycle. n=7; error bar indicates SEM. ANOVA tests were run for an effect on
exosuit assistance (defined as peak exosuit ankle moment). (B) Average collision power, defined as the negative
work performed during the first half of stance divided by stride time, decreased (P=0.002) with increasing
exosuit assistance. (C) Average rebound power, defined as the positive work performed during mid-stance
divided by stride time, decreased (P=5×10-8) with increasing exosuit assistance. (D) Average preload power,
defined as the negative work performed during mid-stance divided by stride time, remained unchanged
(P=0.096) with increasing exosuit assistance. (E) Average push-off power, defined as the positive work
performed during late stance divided by stride time, increased (P=3×10-6) with increasing exosuit assistance.
Fig. S7. Hip and ankle force profiles. Average force profiles for both the ankle (A) and hip (B) over the gait
cycle for each condition.
Fig. S8. Changes in biological joint moment. (A) Biological hip moment normalized to body mass over the gait
cycle for each condition, averaged across participants. Gray shading outlines approximate region during which
assistance was applied. Bar graph on the right indicates the peak biological hip moment during push-off. n=7;
error bar indicates SEM. ANOVA tests were run for an effect on exosuit assistance (defined as peak exosuit
ankle moment). Peak biological hip moment during push-off decreased in magnitude (P=1×10-17) with
increasing exosuit assistance. (B) Biological knee moment assuming no exosuit knee moment. Thus, total and
biological knee moment are equivalent. Peak biological knee moment during push-off decreased in magnitude
(P=0.014) with increasing exosuit assistance. (C) Biological ankle moment. Peak biological ankle moment
during push-off decreased in magnitude (P=3×10-24) with increasing exosuit assistance.
Fig. S9. Changes in biological joint power. (A) Biological hip power normalized to body mass over the gait
cycle for each condition, averaged across participants. Gray shading outlines approximate region during which
assistance was applied. Bar graph on the right indicates the average positive biological hip power and average
negative biological hip power. n=7; error bar indicates SEM. ANOVA tests were run for an effect on exosuit
assistance (defined as peak exosuit ankle moment). Both average positive biological hip power (P=2×10-7) and
average negative biological hip power (P=3×10-9) decreased in magnitude with increasing exosuit assistance.
(B) Biological knee power, assuming no exosuit knee moment. Thus, total and biological knee power are
equivalent. Both average positive biological knee power (P=1×10-4) and average negative biological knee
power (P=5×10-5) decreased in magnitude with increasing exosuit assistance. (C) Biological ankle power.
Average negative biological (total minus exosuit) ankle power decreased in magnitude (P=7×10-6) with
increasing exosuit assistance. Average positive biological ankle power decreased but not significantly
(P=0.099).
Fig. S10. Net biological joint power. Net biological joint power at the hip, knee, and ankle. (A) Average net
biological hip power decreased (P=0.004) with increasing exosuit assistance. (B) Average net biological knee
power did not change (P=0.069) with increasing exosuit assistance. (C) Average net biological ankle power
increased (P=0.006) with increasing exosuit assistance.
P1
P2
P3
P4
P5
P6
P7
MEAN ± SEM
LOW
-0.420
-0.251
+0.206
-0.045
+0.048
-0.108
-0.540
-0.159 ± 0.099
MED
-0.792
-0.233
+0.238
-0.407
-0.324
-0.013
-0.343
-0.268 ± 0.122
HIGH
-1.059
-0.435
-0.611
-0.770
-0.825
-0.546
-0.815
-0.723 ± 0.079
MAX
-1.151
-0.739
-1.399
-1.060
-0.735
-0.931
-1.730
-1.107 ± 0.136
Table S1. Changes in net metabolic rate for each participant. Changes in net metabolic rate of the four active
conditions for each of the seven participants (P1 to P7) in W kg-1 as well as the average across all participants.
Net metabolic reduction was calculated by first subtracting standing metabolics off of the average metabolic
rate from the last two minutes of each condition and then taking the difference in net metabolic rates between
each active condition and the powered-off condition that was recorded within the same 15-min bout of walking.
All rates were normalized by participant’s body mass.