Muhammad Ihsan Hanif

081811733035

Analisis Aktuator pada Assistive dan Rehabilitation Device

Sumber :
- Assistive Device :

Thalman, C. M., Hsu, J., Snyder, L., & Polygerinos, P. (2019). Design of a Soft
Ankle-Foot Orthosis Exosuit for Foot Drop Assistance. 2019 International Conference
on Robotics and Automation (ICRA). doi:10.1109/icra.2019.8794005

- Rehabilitation Device :

Zhu, G., Zeng, X., Zhang, M., Xie, S., Meng, W., Huang, X., & Xu, Q. (2016).
Robot-assisted ankle rehabilitation for the treatment of drop foot: A case study.
2016 12th IEEE/ASME International Conference on Mechatronic and Embedded
Systems and Applications (MESA). doi:10.1109/mesa.2016.7587130
2019 International Conference on Robotics and Automation (ICRA)
Palais des congres de Montreal, Montreal, Canada, May 20-24, 2019
Design of a Soft Ankle-Foot Orthosis Exosuit for Foot Drop Assistance
Carly M. Thalman, Student Member, IEEE,, Joshua Hsu, Student Member, IEEE,, Laura Snyder,
and Panagiotis Polygerinos∗ , IEEE Member
Abstract— This paper presents the design of a soft ankle-
foot orthosis (AFO) exosuit to aid natural gait restoration for
individuals suffering from foot drop. The sock-like AFO is
comprised of soft actuators made from fabric-based, thermally-
bonded nylon and designed to be worn over the users shoes.
The system assists dorsiflexion during swing phase of the gait
cycle utilizing a contracting soft actuator, and provides ankle
joint proprioception during stance with a variable stiffness
soft actuator. A computational model is developed using finite
element analysis to optimize the performance characteristics of
the fabric actuators prior to fabrication, maximize contraction,
and minimize overall volume. The dorsiflexion actuator is able
to achieve a linear tensile force of 197N at 200kPa. The variable
stiffness actuator generates up to 1.2Nm of torque at the
same pressure. The computational model and soft AFO are
experimentally validated and with a healthy participant through
kinematic and electromyography studies. When active the AFO
is capable of reducing by 13.3% the activity of the muscle
responsible for ankle dorsiflexion during the swing phase.
I. INTRODUCTION
Foot drop is described as the inability to properly perform
dorsiflexion at the ankle joint, where the foot cannot clear the
ground during the swing phase of the gait cycle [1]–[3]. The
main ankle dorsiflexor muscle is the tiabilis anterior (TA),
and observed weakness in this muscle can result in the foot
catching on the ground [4]. The TA has been shown to assist
in proper foot placement during heel strike during normal
human locomotion and will be the main muscle that the
exosuit in this paper aims to assist [5] [6]. Muscle weakness
leading to foot drop can occur as a result of nerve injury,
muscular atrophy, stroke, or disease [1] [7]. A common form
of rehabilitation used is gait training, which aims to restore a
natural gait cycle. However, it is estimated that nearly 20% of
those in gait rehabilitation are affected by foot drop, which
can ultimately impede the ability to obtain a normal gait
[8] [9]. Commonly prescribed solutions for individuals with
weak or paretic dorsiflexor muscles are ankle-foot orthosis
(AFOs), which are rigid braces that hold the foot in a neutral
position in the sagittal plane to inhibit movement in the DP
direction [1] [2] [10], and allows for proper execution of
the swing phase [3] [11]. AFOs are typically made from
lightweight, rigid plastic materials that brace the ankle joint
to prevent unwanted movement - however most AFO limit
∗ Corresponding author.
C. M. Thalman, J. Hsu, and P. Polygerinos are with the Poly-
technic School, Ira A. Fulton Schools of Engineering, Arizona
State University, Mesa, AZ 85212, USA. cmthalma@asu.edu;
jahsu@asu.edu; polygerinos@asu.edu
L. Snyder MD is with Barrow Neurological Institute at Dignity Health
St. Josephs Hospital and Medical Center, Phoenix, AZ 85013, USA.
Laura.Snyder@barrowbrainandspine.com
978-1-5386-6027-0/19/$31.00 ©2019 IEEE
Fig. 1. Concept of the soft ankle-foot orthosis (AFO) exosuit designed for
assisting in both (1) dorsiflexion and (2) lateral/medial ankle support and
proprioception during walking.
all ankle movement of the paretic side and may not allow
the user to retrain the strength of their foot [2] [12] [13].
This can result in gait abnormalities and inability to achieve
the full ankle motion [2] [14]–[16]. Excessive repetition of
abnormal patterns can result in further injury or discomfort
to the individual, such as back, hip and knee problems, and
increased probability of tripping and falling [13]–[16].
A wide range of wearable robotics studies that address
lower-body assistance and gait rehabilitation have been con-
ducted [12] [14] [17]–[19]. The benefit of using robotics
to address lower limb and ankle assistance is the ability to
achieve and support more complex motions. Robotic AFOs
aim to assist the user in regaining range of motion and control
of the ankle through the entire gait cycle by enabling specific
mechanisms to regulate movement in specified degrees of
freedom. There are several existing methods that use reha-
bilitative robotics to improve ankle stability, however, many
of these approaches can be heavy, complex, and expensive
to implement [14] [17] [20]. The field of soft robotics has
been a fast-growing area of study, producing a number of
wearable assistive solutions for rehabilitative purposes [12]
[19] [21]–[24].
This paper presents a soft, dynamic ankle-foot orthosis
exosuit to assist impaired users in restoring a natural gait for
individuals suffering from foot drop. The approach uses prin-
ciples of soft robotics to provide a lightweight, comfortable
solution that does not rely on complex mechanical features
to achieve a simple motion [25] [26]. This allows the user
to regain control of dorsiflexion in the foot while walking,
all while avoiding interference with the users natural gait
cycle to prevent further injuries. Section II provides a sys-
tem overview of the proposed solution, the system controls
8436
Fig. 2. (a) Layer shows the sock-like design of the soft AFO exosuit,
which allows the device to be worn over the user’s shoe, (b) Illustration of
straps used to secure the soft AFO exosuit in place around the ankle and
shoe, as well as the sensor placement, and (c) placement of actuators on the
exosuit are shown for both the dorsiflexion and variable stiffness actuators.
and the fabrication processes, while Section III introduces
the modeling of individual components. In Section IV, the
performance and functionality of the AFO is evaluated at
the component-level, and section VI illustrates the system
functioning while donned by a user.
II. AFO EXOSUIT DESIGN
The system is comprised of two main sets of soft actuators
to assist the ankle: the dorsiflexion actuator on the front of
the AFO, and the variable stiffness actuators placed on the
lateral and medial sides of the ankle, both as shown in Fig.
2. Each set serves an individual purpose for a different stage
in the gait cycle. The dorsiflexion actuator assists in lifting
the toes up toward the body as shown in Fig. 3. The variable
stiffness actuators give support to the ankle to prevent ankle
buckling during heel strike, and ensure proper proprioception
before planting the foot on the ground. The synchronization
of these sets of actuators work together to properly orient
the ankle through each stage of the gait cycle. The prototype
is comprised of three layers: the extensible fabric sleeve
that covers the user’s shoe, the foam pads that prevent the
slippage of the sleeve against the skin and helps the exosuit
hold its shape, and the actuators which are embedded within
the sleeve (Fig. 2). The base of the sleeve, which covers the
anterior and posterior of the sole of the shoe, are lined with
an inextensible nylon fabric which provides a point to affix
the actuators to the sleeve to avoid slippage.
The soft dynamic AFO exosuit is powered by an on-board
controls system, which uses a portable pump to actuate the
exosuit without the need for a tethered pneumatic supply.
The majority of the weight of the system is allocated to
the battery and pump, which are encased in a fabric-based
belt to comfortably distribute the added weight on the user’s
body while walking [27]. The controls system is comprised
of (a) force sensitive resistor (FSR) sensors set to monitor
the current position of the foot in the gate cycle, and (b)
fluidic pressure sensors to ensure the maximum operating
threshold of the actuators is not exceeded. The system is
powered by an 11.1V LiPo battery pack which powers the
portable system, with up to nearly 2.5 hours of continuous
use of the portable pump. FSR sensors are placed on the
underside of the exosuit to allow for the reading of ground
8437
Fig. 3. Simplified free-body diagrams for the two main functions of the
soft AFO exosuit - swing phase assist with the dorsiflexion actuator (left)
and ankle support and proprioception during stand (right).
Fig. 4. Controller logic based on GRF input from sensors sets A/B and
C/D on the bottom of the exosuit.
reaction forces (GRF). The soft AFO exosuit itself weighs
only 149.7g while the combined weight of the entire system
including the control belt is 1.6kg, with the added weight
from the portable pump and battery system at at waist.
A. System Controls
The soft AFO exosuit uses a portable control system that
allows the user to walk freely with the exosuit without the
hindrance of a tether or power cord. Figure 4 shows the
structure of the controls applied to the hardware located in
the logic control belt. A pair of sensors are embedded in
the anterior side of the AFO (A and B) and another pair is
affixed to the posterior side (C and D) allow the system
to identify where the user is at within the gait cycle by
monitoring the GRF, as shown in Fig. 2. In combination
the sensors identify when the user is in the preswing or heel
strike phase of their gait following the logic flow as outlined
in Fig. 4. The system uses simple bang-bang control logic to
monitor the GRF of the user to detect initial heel strike and
the transition from toe-off to swing. Two fluidic pressure
sensors are located in the controls pouch and monitor the
activation and deactivation of the two actuators sets in the
soft AFO exosuit throughout walking.
B. Fabrication
The actuators are fabricated from 70D (Denier) Ripstop
Nylon coated in Thermoplastic Polyurethane (TPU) (Seattle

Fig. 5. (a) Free body diagram of the dorsiflexion actuator and approximated
measurements of the foot, (b) dorsiflexion actuator with all critical param-
eters labeled post-inflation assuming each unit (n) forms a perfect circle
when inflated, and (c) the deflated length of a single unit of the dorsiflexion
actuator.
Fabrics, Seattle, WA). The use of thermally bonded fabrics
has been previously shown as an effective approach to
fabrication of soft inflatable actuators [28]–[30]. Two layers
of the fabric are stacked with the TPU sides facing inward.
The actuators are sealed using a customized Cartesian CNC
machine retrofitted with a heated soldering iron to trace the
actuator design for a computer-guided thermal bond between
the two fabric layers for an air-tight seal. The soldering
iron is set to 230◦C, which is experimentally determined to
provide leak proof results.
The seams are set to the smallest width of 2.5 ± 0.5mm,
which is experimentally determined to provide leak proof
results. A 35cm diameter embroidery hoop is used to provide
stability, tension, and an elevated medium to ensure that the
fabric lay uniformly across the CNC workspace. One edge
of the actuator is left open to allow access for installation of
the tube fittings. The remaining open side of the actuator is
sealed with a 5mm impulse sealer, completing the fabrication
of the actuators.
III. MODELING AND DESIGN PARAMETERS
A. Theoretical Modeling
In order to determine the functional constraints of the
system, an analytical model of the most critical component
of the system - the dorsiflexion actuator - has been derived
to estimate the needed amount of contraction from the dor-
siflexion actuator. Furthermore, the tension force needed to
actuate the foot in dorsiflexion, while still allowing full range
of motion for plantarflexion when the system is not active,
is also included in the model. The method for identifying
the amount of contraction needed to perform dorsiflexion is
determined using the law of cosines and is denoted as,
 behavior of the soft materials (Fig. 6). Computational mod-
Lapprox (θ f ) = Lh2 + L2f − 2Lh L f cos(θ f ), (1)
8438
where, θ f is there current ankle DP angle and ranges from
95◦ ≤ θ f ≤ 145◦ as shown in Fig. 6(a). The angle at which
the ankle is assumed to be in the natural, relaxed position is
at 115◦ , allowing for 20◦ dorsiflexion and 30◦ plantation. Lh
is the height at which the actuator is mounted on the shin
from the joint, and L f oot is length between the joint and the
metatarsal at which the actuator is affixed for applied force,
as shown in Fig. 6(a). The initial length is assumed to allow
for the full range of plantarflexion, while the final length
assists in the full range of dorsiflexion. The tension force is
assumed to be linearly proportional to the input pressure for
the purposes of the following force model,
F = L f sin(φ ), (2)
where, φ is the angle between the F vector and its vertical
component. The angle is obtained assuming the foot has
an upward slope of 25◦ in the neutral position, which is
subtracted off the right angle produced between the ground
and the vertical force vector. The resulting maximum force
to achieve the desired contraction percentage for the actuator
is estimated as,
Li − L f
ε= , (3)
Li
L(x1 ) = (n − 1)d + nx1 + 2d, (4)
where Li is the initial length of the actuator prior to
inflation (Fig. 6c), and L f is the final length after contraction
(Fig. 6b), as modeled similarly in previous works [31]. The
distance between each unit is denoted as d, n is the number
of units and x1 is the initial height of the deflated unit. In
order to predict the final length of the dorsiflexion actuator,
the contraction is estimated as follows:

| x1 | −2(cos(2θ1 ) − 1)
L f (θ1 ) = , (5)
2
2
sin−1 ( ) ≤ θ1 ≤ 90◦ , (6)
π
where, θ1 is the half angle between the center line of
the unit and the edge, as shown in Fig. 6(b). The values of
θ1 ranges from where the bounds are set at the points at
which the unit forms a perfect circle, or becomes perfectly
flat. This model bases the final contraction percentage on
the input constants d and x1 . Using the specific dimensions
of the dorsiflexion actuator of d = 0.02cm and x1 = 2.5cm
the final contraction ratio is estimated at 36.5% of the initial
length, assuming the units can inflate to perfect circles. This
value is comparable to the values achieved experimentally
and through FEA optimization, as it is further discussed in
the following section.
B. Finite Element Analysis (FEA) of Soft Fabric Materials
Prior to fabrication, the contraction of the dorsiflexion ac-
tuator is characterized using a finite element analysis (FEA)
package (ABAQUS, Dassault Systems, Vlizy-Villacoublay,
France) to monitor a computer-generated simulation of the
eling of soft materials has been shown in previous work to
Fig. 6. (a) Deflated state the single two-unit dorsiflexion actuator, (b) two-
unit dorsiflexion when inflated, and (c) a side few of the inflated actuator
to show contraction and comparison to idealized model.
Fig. 7. Results of 50 FEA simulations showing the contraction percentage
of the dorsiflexion actuator varying the height (x1 ) of each unit as a ratio of
the width (a), and varying the side gaps (x2 ) as a ratio of the overall width
(b).
provide an effective means to understanding the behavior and
performance of soft actuators [32] [33].
The models are created using two overlaid 2D homoge-
neous shell bodies in the rectangular shape of the actuator.
The seams are created by implementing tie constraints as
denoted by the shaded regions in Fig. 6, and the actuator
is fixed along the top seam to prevent movement in the
top section during inflation. A uniform pressure force is
applied to the internal faces of the two shells to behave
as a pneumatic pressure source. The material properties
are represented through an elastic behavior, with a Youngs
modulus of 224MPa, identified in former studies with TPU
coated nylon fabrics [32]. During inflation, the thin shells
move and crumple as the TPU coated nylon fabric does
in reality, and is observed to mimic the behavior of the
inflatable fabrics selected. The dimensions of the actuators
are optimized to maximize contraction percentage while
minimizing the internal volume of the actuator. The con-
traction is recorded through the results of the FEA while
varying three parameters individually, and holding all other
parameters constant: the ratio of the height to the width of
each unit (x1 ), the gap on either side of the seam connecting
each unit (x2 ) and the internal operating pressure of the
actuator (P). It is assumed that these three parameters will
have the greatest impact on the overall contraction ratio and
are varying sequentially as specified.
A script is developed in python to automate the FEA pro-
cess, allowing for easy variation of each parameter through
every iteration of the dimensional analysis of the dorsiflexion
8439
Fig. 8. (a) FEA of the dorsiflexion actuator, showing displacement between
P = atm and P = 200kPa and (b) real fabricated actuator comparison when
inflated to 200 kPa)
actuator with the width fixed and the other parameters are
varied as ratios of a fixed width. This is done for the total
length of the actuator,
f (x) = 2d + (n − 1)d + nx1 (7)
where d is the width of the seam, n is the number of
pleats, and x1 is the height as a ratio of the width, as seen
in Fig. 6. The parameter x1 is evaluated; the height/width
ratio of each unit with all other parameters held constant.
First, x1 is varied from 0.1 : 1 to a 1 : 1 ratio. A decline in
the contraction percentage from Fig. 7(a) is recorded from
the results of the FEA simulations. The peak contraction
percentage is achieved at a ratio of 0.5:1 height/width ratio.
The gap size is evaluated as a ratio of the width given the
following equation,
g(x) = W /x2 , (8)
where, W is the width of the actuator and x2 is the specified
ratio of the width to the gap size that is varied for each
iteration. The gap/width ratio is evaluated from 1:1 to 6:1,
at which point the tolerances achievable given the current
manufacturing processes is no longer feasible. Figure 7(b)
shows the maximum contraction obtained gap, g, of 5.3/1
W /x2 ratio. However, due to physical limitations such as
inflation time of the actuator, this gap size is increased to
a ratio of 4:1 to allow for better airflow with the particular
width selected.
TABLE I
PARAMETER S ELECTION FOR D ORSIFLEXION ACTUATOR
Parameter Ratio Chosen Selected Dimension
W 1 (base) 5cm
d Fixed 2mm
n N/A 8
x1 0.5:1 2.5cm
x2 4:1 0.625cm
The internal actuator pressure, P, peaks the contraction
at 230kPa (Fig. 8). Reducing the operating pressure means
more efficient operation, as reducing the needed pressure
decreases inflation/deflation time of the actuator. As a result
of the FEM optimization, the parameters are set as follows:
x1 = 0.5W , x2 = W /4, and an operating pressure with a range
of 150kPa to 200kPa. Figure 8 shows a discrepancy of 7.9%
between FEA and experimentally measured contraction of
the dorsiflexion actuator. This error could be due to the
oscillations seen in the materials during dynamic explicit
simulation conditions, as well as minor torsion observed on
the bench during inflation of the actuators at higher pressure,
adding to overall contraction.
IV. EVALUATION OF SOFT COMPONENTS
A. Dorsiflexion Actuator
The dorsiflexion actuator is evaluated following the same
testing protocol used in testing Mckibben actuators [34] Fig. 9. (a) Quasistatic isometric test and (b) constant pressure test
[35], as the primary function is consistent between the two performed for the dorsiflexion actuator, repeated cyclically for a total of
designs: uniaxial contraction. There have been several other three trials. Max force of 197.1N found with a standard deviation of 2.2N.
successful applications of inflatable fabric-based dorsiflexion
actuators [31]. Two tests are performed as shown in Fig. 9.
A quasistatic isometric test is performed using a universal
tensile testing machine (UTM) (Instron 5944, Instron Corp.,
High Wycombe, United Kingdom), where both ends of the
dorsiflexion actuator are clamped and held at a fixed position.
The pressure is increased in increments of 10kPa from 0kPa
up to the maximum operating pressure of 200kPa, and back
down to 0kPa. This is performed cyclically for three trials,
with the UTM being reset and recalibrated between each
trial. The dorsiflexion actuator is able to generate a maximum
of 197.1 ± 2.2N when pressurized up to the maximum
operating pressure of 200kPa. Fig. 10. (a) Constant pressure test performed with various applied loads at
the tip of the cantilever beam, and (b) a constant load test with incrementally
A constant pressure test is performed with a dorsiflexion increased pressure values
actuator on the UTM, with pressure held constant at the
maximum of 200kPa throughout the entirety of the test. The measured each time load is added, and this is repeated for
dorsiflexion actuator is clamped at both ends, with a load cell three iterations. It is found that the actuator can stay under
affixed to the top end. At the start of the test, the top clamp 10 ± 0.17 ◦ deflection with a load of 1Kg applied at the tip.
is released and allowed to move freely downward to measure The variable stiffness actuator is also assessed with constant
the contraction of the actuator, as well as the pulling force load with varying pressures. The system is pre-loaded with
generated. The clamp is released from its position, moves 1kg, the equivalent of 1.2Nm of torque with the actuator
downward until the force measured reads 0N, then moves fully deflated. This is roughly 12% of the maximum ankle
back to its original fixed position. This is repeated cyclically torque during rapid buckling The variable stiffness actuator
for three trials for the soft actuator. A total displacement of is then inflated in increments of 10kPa up to 150kPa with
43.1mm ± 0.5mm is observed, or a contraction of 32.3%, the constant load still in effect. A deflection from a fully
compared to the 36.5% predicted assuming ideal contraction bend actuator at 90◦ to 12.0 ± 0.46 ◦ state of deflection is
of each unit to form a perfect circle. observed.
B. Variable Stiffness Actuators V. USER TESTING AND EVALUATION
The variable stiffness actuators are evaluated based on In order to test the efficacy of the AFO exosuit, a healthy
the ability to resist the lateral and medial movement of the participant (gender: female, age: 23 years old, height: 1.67m,
ankle, which can lead to excessive supination or pronation weight: 47.6kg,) is used to validate the performance of the
of the ankle and cause buckling [36]. A single variable design with preliminary participant testing results. Surface
stiffness actuator is placed in a custom vice with half of electromyography (sEMG) sensors (Delsys Trigno, Delsys,
the actuator length fixed in place and the other free-floating Natick, MA) is used to track the activity of the participant’s
as a cantilever beam as shown in Fig 10. This mimics the tibialis anterior muscle, as well as the movement of the
placement of the variable stiffness actuators on the ankle, participant’s ankle throughout the test with an sEMG sensor.
fixed at one end by the soft AFO exosuit straps and the other The participant is asked to walk on a treadmill at a controlled
free to bend and move against the calcaneus of the ankle. speed of 0.4m/s for a total of 3 minutes. This is performed
The variable stiffness actuator is evaluated under constant for three trials, with 30 second rest intervals interjected
pressure of 150kPa, with increasing load applied at the end between each trial. The exosuit functions so that a healthy
of the cantilevered end. The angle of actuator deflection is participant without foot drop experiences reduced muscle

8440
activity in the tibialis anterior during the swing phase of
the gait cycle.
The maximum voluntary contraction (MVC) and the mus-
cle activity during relaxation is collected as per standard
International Society of Electrophysiology and Kinesiology
(ISEK) protocols [37]. A baseline resting activity recorded
with the participant sitting at rest in a chair, and MVC is
also recorded with the participant exerting maximum effort
in dorsiflexion against a fixed surface. The system records
the pressure to each actuator at the logic control belt and the
GRFs during preswing and heel strike through the controller.
The data is synchronized with that of the sEMG sensors and
ankle joint angle though the motion capture data to generate
the gait profile averaged across three full gait cycles. This is
performed first without the user wearing the exosuit, then
wearing the inactive exosuit, and finally with the exosuit
active. Six steps are averaged for the inactive and active
exosuit data, and a 13.3% decrease in muscle activity is
observed during dorsiflexion when isolating the swing phase
of the gait from 60% to 95% . These preliminary results
indicate that the exosuit is able to reduce muscle effort in the
main muscle related to dorsiflexion during the swing phase
in a healthy participant.
VI. CONCLUSIONS AND DISCUSSION
In this paper, the design, characterization and evaluation
of a soft, lightweight AFO exosuit to assist foot drop was
presented. The soft AFO exosuit was designed to assist the
Fig. 11. State of the system during a single gait cycle from heel strike
wearer in achieving the movement of dorsiflexion during to heel strike. (a) show the output of the fluidic pressure sensors indicating
walking, as well as providing support to the ankle during heel when each soft actuator is active, b) GRF readings indicating part of the
strike. The exosuit is easily donned/doffed over the user’s foot is on the ground, and c) sEMG data normalized over MVC for the
tibialis anterior muscle. The mean values of muscle activity are shown for a
athletic shoe and works off of a fully portable system. The single user with exosuit active and without wearing the exosuit. The average
actuators are made from a TPU-coated nylon fabric, which is difference of the exosuit being worn while inactive is subtracted from the
heat sealed to allowed for pneumatic actuation. A contracting active plot.
actuator for dorsiflexion is placed on the front of the foot to Future work will include exploring methods for increas-
pull the toes upward at preswing during gait, and the variable ing the current speed of actuation and allowing for faster
stiffness actuators inflate at heel strike to provide stability for actuation to support walking speeds faster than the 0.4m/s
the ankle before transitioning to terminal stance. tested. New controls techniques will be evaluated to allow the
An FEA study is performed to optimize the contraction exosuit to more easily adapt to different user’s gait patterns.
percentage of the dorsiflexion actuator, while minimizing While testing with a healthy participant showed reduction
volume. The actuator dimensions are selected by varying in muscle activity in the tibialis anterior, the sample size of
individual parameters while holding all others constant, until healthy participants will be increased in future studies. After
contraction is maximized. The actuators are then tested on a obtaining more extensive preliminary results with a larger
UTM for uniaxially force under contact pressure and constant sample size, the soft AFO exosuit will also be tested for
position. The dorsiflexion actuators are able to generate 197.1 its ability to assist impaired participants suffering from foot
± 2.2N and contract a total of 31%. The variable stiffness drop.
actuators are evaluated for the ability to withstand laterally ACKNOWLEDGMENTS
applied forces and tested as cantilever beams, demonstrating C.M. Thalman is funded by the National Science Foun-
their ability to withstand a torque applied quasi-statically up dation Graduate Research Fellowships Program. The authors
to 1.2Nm while staying within 10 ± 0.17 ◦ deflection from would like to thank Caleb Carlson and Kayleigh Gavin for
their neutral position. A healthy participant was recruited to their contributions. The research is partially supported by the
preliminary test the soft AFO exosuit during walking using Arizona Department of Health Services - New investigator
the on-board control hardware. The maximum walking speed Award (ADHS18-198863).
of 0.4m/s was performed while actuator pressure, GRF, and R EFERENCES
tibialis anterior muscle activity were recorded. An average of
[1] P. M Kluding, K. Dunning, M. W Odell, S. S Wu, J. Ginosian, J. Feld,
13.3% reduction was observed sin the TA during the swing and K. McBride. Foot drop stimulation versus ankle foot orthosis after
phase. stroke: 30-week outcomes. Stroke, 44(6):1660–1669, 2013.

8441
 [2] J F Geboers, M R Drost, F Spaans, H Kuipers, and H A Seelen.
Immediate and long-term effects of ankle-foot orthosis on muscle
activity during walking: a randomized study of patients with unilateral
foot drop. Archives of physical medicine and rehabilitation, 83(2):240–
245, 2002.
[3] R. Chin, E. T Hsiao-Wecksler, E. Loth, G. Kogler, S. D Manwaring,
S. N Tyson, K A Shorter, and J N Gilmer. A pneumatic power
harvesting ankle-foot orthosis to prevent foot-drop. Journal of neuro-
engineering and rehabilitation, 6(1):19, 2009.
[4] F D Conlin, P G Johnson, and John E Sinning Jr. The etiology and
repair of rotary ankle instability. Foot & ankle, 10(3):152–155, 1989.
[5] HI Dubo, M. Peat, DA Winter, AO Quanbury, DA Hobson, T Steinke,
and G Reimer. Electromyographic temporal analysis of gait: normal
human locomotion. Archives of physical medicine and rehabilitation,
57(9):415–420, 1976.
[6] J Hamill and K M Knutzen. Biomechanical basis of human movement.
Lippincott Williams & Wilkins, 2006.
[7] R B Stein, D G Everaert, A K Thompson, S Chong, M Whittaker,
J Robertson, and G Kuether. Long-term therapeutic and orthotic effects
of a foot drop stimulator on walking performance in progressive and
nonprogressive neurological disorders. Neurorehabilitation and neural
repair, 24(2):152–167, 2010.
[8] Y Laufer, J M Hausdorff, and H Ring. Effects of a foot drop
neuroprosthesis on functional abilities, social participation, and gait
velocity. American journal of physical medicine & rehabilitation,
88(1):14–20, 2009.
[9] D G Everaert, R B Stein, G M Abrams, A W Dromerick, G E
Francisco, B J Hafner, T N Huskey, M C Munin, K J Nolan, and
C V Kufta. Effect of a foot-drop stimulator and ankle–foot orthosis on
walking performance after stroke: a multicenter randomized controlled
trial. Neurorehabilitation and neural repair, 27(7):579–591, 2013.
[10] J. S Jaivin, J. O Bishop, W G Braly, and H S Tullos. Management of
acquired adult dropfoot. Foot & ankle, 13(2):98–104, 1992.
[11] C. E. Buckon, S. S. Thomas, S. Jakobson-Huston, M. Moor, M. Suss-
man, and M. Aiona. Comparison of three ankle–foot orthosis config-
urations for children with spastic diplegia. Developmental Medicine
and Child Neurology, 46(9):590–598, 2004.
[12] P Malcolm, S Lee, S Crea, C Siviy, F Saucedo, I Galiana, F A
Panizzolo, K G Holt, and C J Walsh. Varying negative work assistance
at the ankle with a soft exosuit during loaded walking. Journal of
neuroengineering and rehabilitation, 14(1):62, 2017.
[13] I. Wiszomirska, M. Błażkiewicz, K. Kaczmarczyk, G. Brzuszkiewicz-
Kuźmicka, and A Wit. Effect of drop foot on spatiotemporal,
kinematic, and kinetic parameters during gait. Applied bionics and
biomechanics, 2017, 2017.
[14] J. Blaya and H. Herr. Adaptive control of a variable-impedance ankle-
foot orthosis to assist drop-foot gait. IEEE Transactions on neural
systems and rehabilitation engineering, 12(1):24–31, 2004.
[15] J. F Lehmann, S. M Condon, B. J de Lateur, and J C Smith. Ankle-foot
orthoses: effect on gait abnormalities in tibial nerve paralysis. Archives
of physical medicine and rehabilitation, 66(4):212–218, 1985.
[16] Jr MT Balmaseda, SH Koozekanani, MT Fatehi, C Gordon, PH Drey-
fuss, and EC Tanbonliong. Ground reaction forces, center of pressure,
and duration of stance with and without an ankle-foot orthosis.
Archives of physical medicine and rehabilitation, 69(12):1009–1012,
1988.
[17] Y. D. Li and E. Hsiao-Wecksler. Gait mode recognition using an
inertial measurement unit to control an ankle-foot orthosis during
stair ascent and descent. In ASME 2012 5th Annual Dynamic
Systems and Control Conference joint with the JSME 2012 11th
Motion and Vibration Conference, pages 743–752. American Society
of Mechanical Engineers, 2012.
[18] A T Asbeck, R J Dyer, A F Larusson, and C J Walsh. Biologically-
inspired soft exosuit. In Rehabilitation robotics (ICORR), 2013 IEEE
international conference on, pages 1–8. IEEE, 2013.
[19] D.G. Caldwell, N. G. Tsagarakis, S. Kousidou, N. Costa, and
I. Sarakoglou. ”Soft” Exoskeletons for Upper and Lower Body
Rehabilitation Design, Control and Testing. International Journal
of Humanoid Robotics, 04(03):549–573, 2007.
[20] Y Park, B Chen, N O Pérez-Arancibia, D Young, L Stirling, R J Wood,
E C Goldfield, and R Nagpal. Design and control of a bio-inspired soft
wearable robotic device for ankle–foot rehabilitation. Bioinspiration
& biomimetics, 9(1):016007, 2014.
[21] S. Sridar, P. H. Nguyen, M. Zhu, Q. P. Lam, and P. Polygerinos.
Development of a soft-inflatable exosuit for knee rehabilitation. In
8442
2017 IEEE/RSJ International Conference on Intelligent Robots and
Systems (IROS), pages 3722–3727, Sept 2017.
[22] M. Xiloyannis, L. Cappello, B. Khanh Dinh, C. W. Antuvan, and
L. Masia. Design and preliminary testing of a soft exosuit for assisting
elbow movements and hand grasping. Biosystems and Biorobotics,
15:557–561, 2017.
[23] P. Polygerinos, K.C. Galloway, S. Sanan, M. Herman, and C.J.
Walsh. EMG Controlled Soft Robotic Glove for Assistance During
Activities of Daily Living. 2015 IEEE International Conference on
Rehabilitation Robotics (ICORR), pages 55–60, aug 2015.
[24] C. M. Thalman, Q. P. Lam, P. H. Nguyen, S. Sridar, and P. Polygerinos.
A Novel Soft Elbow Exosuit to Supplement Bicep Lifting Capacity.
2018 IEEE/RSJ International Conference on Intelligent Robots and
Systems, IROS, 2018.
[25] M. Yandell, B. Quinlivan, D. Popov, C. Walsh, and K. Zelik. Physical
interface dynamics alter how robotic exosuits augment human move-
ment: implications for optimizing wearable assistive devices. Journal
of neuroengineering and rehabilitation, 14(1):40, 2017.
[26] P. Polygerinos, Z. Wang, R. Galloway, K. Wood, and C Walsh. Soft
robotic glove for combined assistance and at-home rehabilitation.
Robotics and Autonomous Systems, 73:135–143, 2015.
[27] R. Browning, J. R. Modica, R. Kram, and A. Goswami. The effects
of adding mass to the legs on the energetics and biomechanics of
walking. Medicine & Science in Sports & Exercise, 39(3):515–525,
2007.
[28] P. Polygerinos, N. Correll, S.A. Morin, B. Mosadegh, C.D. Onal,
K. Petersen, M. Cianchetti, M.T. Tolley, and R.F. Shepherd. Soft
Robotics: Review of Fluid-Driven Intrinsically Soft Devices; Manu-
facturing, Sensing, Control, and Applications in Human-Robot Inter-
action. Advanced Engineering Materials, pages e201700016—-n/a.
[29] R. F. Natividad, M. R. Del Rosario, P. C. Y. Chen, and C. H. Yeow. A
hybrid plastic-fabric soft bending actuator with reconfigurable bending
profiles. In 2017 IEEE International Conference on Robotics and
Automation (ICRA), pages 6700–6705, May 2017.
[30] T.H. Koh, N. Cheng, H.K. Yap, and C.H. Yeow. Design of a Soft
Robotic Elbow Sleeve with Passive and Intent-Controlled Actuation.
Frontiers in Neuroscience, 11(October):597, 2017.
[31] R. Niiyama, D. Rus, and S. Kim. Pouch motors: Printable/inflatable
soft actuators for robotics. In Robotics and Automation (ICRA), 2014
IEEE International Conference on, pages 6332–6337. IEEE, 2014.
[32] W. Adams, S. Sridar, C. M. Thalman, B. Copenhaver, H. Elsaad, and
P. Polygerinos. Water pipe robot utilizing soft inflatable actuators.
In 2018 IEEE International Conference on Soft Robotics (RoboSoft).
IEEE, 2018.
[33] B. Mosadegh, P. Polygerinos, C. Keplinger, S. Wennstedt, R.F. Shep-
herd, U. Gupta, J. Shim, K. Bertoldi, C.J. Walsh, and G.M. Whitesides.
Pneumatic networks for soft robotics that actuate rapidly. Advanced
functional materials, 24(15):2163–2170, 2014.
[34] C. Chou and B. Hannaford. Measurement and modeling of mckibben
pneumatic artificial muscles. IEEE Transactions on robotics and
automation, 12(1):90–102, 1996.
[35] M. Doumit, A. Fahim, and M. Munro. Analytical modeling and
experimental validation of the braided pneumatic muscle. IEEE
transactions on robotics, 25(6):1282–1291, 2009.
[36] G B Wilkerson. Biomechanical and neuromuscular effects of ankle
taping and bracing. Journal of Athletic training, 37(4):436, 2002.
[37] Standards for reporting emg data. Journal of Electromyography and
Kinesiology, 38:I – II, 2018. Neuromechanics of fine hand-motor tasks.
Muhammad Ihsan Hanif
081811733035
Assistive Device

Terdapat 2 aktuator yang dipakai, salah satunya adalah modifikasi dari pneumatic actuator, yakni
soft actuator. Pemilihan aktuator ini ditujukan untuk melakukan gerakan dorsifleksi pada kaki
(ujung kaki diangkat keatas) sehingga bisa digunakan untuk membantu penderita drop foot untuk
berjalan dengan normal. Terdapat juga Variable Stiffness Actuator yang merupakan modifikasi
dari motor untuk menjaga pergelangan kaki agar tidak bergoyang ketika berjalan karena berat
dari AFO. Kedua aktuator tersebut bekerja pada tiap tahapann pada siklus gait.

Alat ini bekerja dengan mengambil data dari 4 sensor yang terletak di bagian bawah alas untuk
mengetahui fase pada siklus gait dengan mengetahui GRF dari pasien. Fase tersebut yang
kemudian menentukan bagaimana aktuator akan bekerja pada setiap fase.

Secara umum, desain AFO ini cukup portable sehingga bisa digunakan sebagai Assistive Device
yang layak dipakai sehari hari untuk penderita drop foot. Alat ini mampu memberikan 1 DOF
pada kaki (dorsifleksi) tanpa bisa melakukan gerakan inversi atau eversi.
 Robot-Assisted Ankle Rehabilitation for the
Treatment of Drop Foot: A Case Study
Mingming Zhang, Shane Xie*, Wei Meng
Department of Mechanical Engineering
The University of Auckland
Auckland, New Zealand
* s.xie@auckland.ac.nz
Guoli Zhu, Xiangfeng Zeng
The State Key Laboratory of Digital Manufacturing
Equipment and Technology
Huazhong University of Science & Technology
Wuhan, China
Abstract— This paper involves the use of an intrinsically-
compliant ankle rehabilitation robot for the treatment of drop
foot. The robot has a bio-inspired design by employing four Festo
fluidic actuators that mimic skeletal muscles to actuate three
rotational degrees of freedom (DOFs). A position controller in task
space was developed to track the predefined trajectory of the end
effector. The position tracking was achieved by the length tracking
of each actuator in joint space by inverse kinematics. A stroke
patient with drop foot participated in the trial as a case study to
evaluate the potential of this robot for clinical applications. The
patient gave positive feedback in using the ankle robot for the
treatment of drop foot, although some limitations exist. The
trajectory tracking showed satisfactory accuracy throughout the
whole training with varying ranges of motion, with the root mean
square deviation (RMSD) value being 0.0408 rad and the
normalized root mean square deviation (NRMSD) value being
8.16%. To summarize, preliminary findings support the potential
of the ankle rehabilitation robot for clinical applications. Future
work will investigate the effectiveness of the robot for treating
drop foot on a large sample of subjects.
Keywords—robot-assisted; ankle; rehabilitation; drop foot
I. INTRODUCTION
Drop foot is very common following neurological injuries,
such as stroke and spinal cord injury (SCI) [1, 2]. Based on an
up-to-date report from the American Heart Association,
approximately 795,000 people experience a new or recurrent
stroke (ischemic or hemorrhagic) in the United States each year,
of which about 610,000 are the first events and the remainder
are recurrent events [3]. An estimated 60,000 stroke survivors
live in New Zealand [4], and around 3,000 stroke patients are
discharged from hospitals each year with significantly abnormal
gait pattern [5]. In New Zealand, every year approximately 80 to
130 people are diagnosed with spinal cord impairment through
injury or medical causes [6]. Many of these neurologically
impaired subjects have the symptom of drop foot, which affects
The University of Auckland, Faculty of Engineering Research Development
Fund 3625057 (Physical Robot-Human Interaction for Performance-Based
Progressive Robot-Assisted Therapy).
978-1-5090-6190-7/16/$31.00 ©2016 IEEE
Xiaolin Huang, Qun Xu
Rehabilitation Department
Tongji Hospital
Wuhan, China
their lives and those of many others, especially their families.
Drop foot prevents them from lifting their feet and toes properly
when walking, affecting the balance, general mobility, and self-
confidence. Walking like this is slow, uncomfortable and tiring,
taking great effort and concentration, and it also leads to hip,
pelvis and back pain.
Treatments of drop foot are variable depending on specific
causes. While treatments, such as braces and orthotics [7-9],
functional electrical stimulation [10-12] and surgery [13], have
been demonstrated to be effective for drop foot, physiotherapy
as the primary treatment is commonly prescribed together with
other options such as functional electrical stimulation to
maximize the function of the patient [14, 15]. Strengthening
exercises of the muscles within the foot and the lower limbs help
maintain muscle tone, and improve gait pattern associated with
drop foot. For the treatment of drop foot, joint stretching along
dorsiflexion is important and requires large driven torque from
the robot. A conventional physiotherapy treatment of drop foot
usually requires cooperative and intensive efforts from both
therapists and patients over prolonged sessions [17].
Robot-assisted ankle rehabilitation solutions, as therapeutic
adjuncts to facilitate clinical practice, have been actively
researched during the past few decades. The robot could also
provide a rich stream of data using intelligent sensing units to
facilitate patient diagnosis, customization of the therapy, and
maintenance of patient records. There are two types of ankle
rehabilitation devices. In one group are wearable exoskeletons,
such as the MIT Anklebot developed by Roy et al. [18] and the
bio-inspired soft ankle robotic device developed by Park et al.
[19]. The other group consists of various platform-based robots.
These robots usually have a fixed platform and a moving one
[20-24]. While Zhang et al. [25] demonstrated the effectiveness
of existing rehabilitation robots in reducing ankle impairments
caused by neurological injuries, most of them suffer from a
variety of limitations when used for the treatment of drop foot.
Exoskeleton devices focus more on gait training rather than only
ankle exercises [18, 19], which makes them unsuitable for direct
treatment of drop foot. Some parallel robots with misaligned
rotation centers with ankle joints are also unsuitable for this
application [26-28]. The training using these devices requires
synergic movement of the lower limb from the patient. In
contrast, the manipulator wherein the rotation center of the robot
coincides with the ankle joint can be considered to be suitable
for the treatment of drop foot. This kind of robots can have a
single range of motion (DOF) or multiple DOFs. Zhang et al.
[20] presented a single-DOF ankle robot for joint stretching and
its efficacy has been demonstrated on patients with spasticity or
contracture. Two parallel robots [21, 22] have not been clinically
evaluated due to the lack of enough driven torque.
A new ankle rehabilitation robot was recently developed
using four Festo Fluidic muscles (FFMs) in our group. It has
three rotational DOFs that are ankle dorsiflexion/plantarflexion,
inversion/eversion, and adduction/abduction, respectively. This
robot has been significantly improved with respect to previous
prototypes [21, 22]. Its advantages include the use of compliant
actuators, three DOFs for three-dimensional ankle training, and
large driving torque when used for joint stretching and muscle
strengthening. These features make its applications more
extensive with respect to other ankle rehabilitation robots. While
this robot can be used for the treatment of drop foot due to the
large generation capacity of driving torque, its use and efficacy
have not evaluated yet. This paper will investigate and evaluate
the use of this ankle robot on neurologically impaired subjects
with drop foot.
II. METHODS
A. Ankle Rehabilitation Robot
The ankle rehabilitation robot has three rotational DOFs, see
shown in Fig1. It has a bio-inspired design by mimicking the
configuration and actuation of the ankle joint by natural muscles.
Thus this robot is actuated using four FFMs (FESTO DMSP-20-
400N) in parallel. Four proportional pressure regulators (FESTO
VPPM-6L-L-1-G18-0L6H) are used for the pressure control of
individual FFM. The robot, as a parallel mechanism, consists of
a fixed platform and a moving platform, of which the moving
one is actually a three-link serial manipulator with three
rotational DOFs.
Fig. 1. An intrinsically-compliant ankle rehabilitation robot with three DOFs.
(DP: dorsiflexion/plantarflexion; IE: inversion/eversion; and AA:
adduction/abduction).
In this robot, three magnetic rotary encoders (AMS
AS5048A) are installed along each axis for measuring three-
dimensional angular positions of the footplate and the human
ankle. It is assumed here that there is no relative motion between
the footplate and the human foot during the training, thus the
measured position of the foot plate equals that of the involved
foot. There are four single-axis load cells (FUTEK LCM 300)
for measuring contraction forces of four FFMs, and a six-axis
load cell (SRI M3715C) for the measurement of real-time
human-robot interaction forces and torques. These electronic
components communicate with an embedded controller (NI
Compact RIO-9022). The six-axis load cell communicates with
the controller through the RS232.
B. Muscle Length Control in Joint Space
The trajectory control of the end effector is required to
implement passive or active training on a rehabilitation robot.
The position control of this ankle rehabilitation robot can be
achieved by controlling individual FFM length in joint space, as
shown in Fig.2. The desired individual FFM length is calculated
by inverse kinematics based on the desired position of the end
effector, while, as the feedback to the proportional–integral–
derivative controller (PID) controller, the actual individual FFM
length is obtained by inverse kinematics based on the measured
position of the end effector. This joint space position controller
outputs four pressure values that directly go to four proportional
pressure regulators for the actuation of the robot.
Specifically, the desired trajectory can be predefined by a
physiotherapist and denoted as Ʌୢ ሺ–ሻ in (1). The measured
trajectory is obtained from three magnetic rotary encoders and
denoted as Ʌ୫ ሺ–ሻ in (1). Individual FFM length can be
calculated using (2) based on inverse kinematics and AARR
configuration, where Žୢସൈଵ ሺ–ሻ and Ž୫
ସൈଵ ሺ–ሻ respectively represent
desired and measured FFM lengths, Ɋ is a coefficient that relates
the FFM length to the link length and depends on the AARR
configuration, Յସൈଷ relates the link length to the position of the
robotic end effector and depends on the inverse kinematics of
the AARR. Lastly, the error ‡ସൈଵ ሺ–ሻ shown in (3) is input to the
PID controller, and the desired individual FFM pressure can be
calculated according to (4) with well-tuned  ୮ ,  ୧ , and  ୢ .
Ʌ ሺ–ሻ ൌ ሾɅୢୈ୔ ሺ–ሻ Ʌୢ୍୉ ሺ–ሻ Ʌୢ୅୅ ሺ–ሻሿ୘
ቊ ୢ ୫ (1)
Ʌ୫ ሺ–ሻ ൌ ሾɅ୫ ୫
ୈ୔ ሺ–ሻ Ʌ୍୉ ሺ–ሻ Ʌ୅୅ ሺ–ሻሿ
୘
Žୢସൈଵ ሺ–ሻ ൌ ɊՅସൈଷ Ʌୢ ሺ–ሻ
൜୫ (2)
Žସൈଵ ሺ–ሻ ൌ ɊՅସൈଷ Ʌ୫ ሺ–ሻ
‡ସൈଵ ሺ–ሻ ൌ Žୢସൈଵ ሺ–ሻ െ Ž୫
ସൈଵ ሺ–ሻ (3)
୲
’ସൈଵ ሺ–ሻ ൌ  ୮ ‡ସൈଵ ሺ–ሻ ൅  ୧ න ‡ସൈଵ ሺ–ሻ†– ൅
଴
(4)
†‡ସൈଵ ሺ–ሻ
ୢ
†–
Fig. 2. The flow chart of individual muscle length control in joint space. (PID: proportional–integral–derivative controller)

C. Participant and Training Protocol
A subject (male, 68 years, six months post stroke) with drop
foot on the left participated in this trial as a preliminary study.
This participant can follow the instruction during the training,
and communicate well with the physiotherapist. The subject
gave written consent to participate in the trial. This ethics
approval was obtained from the University of Auckland, Human
Participants Ethics Committee (011904).
increased to 0.15 rad. It was further increased to 0.2 rad after the
moment of the 200th second, when the patient felt slightly tight
at his ankle joint. The robot kept this range of motion for the
training during the period of 200 to 725 seconds. As the patient
required, the amplitude of the training trajectory was finally
adjusted to 0.25 rad, when the patient felt obvious ankle
stretching. The whole training lasted about 15 minutes (900
seconds).

Although this robot is developed with three rotational DOFs
(including ankle dorsiflexion/plantarflexion, inversion/eversion,
and adduction/abduction), training therapy is solely conducted
along dorsiflexion and plantarflexion where patients with drop
foot usually have difficulties in lifting their toes. Before robot-
assisted ankle training, a preliminary assessment was conducted
by a physiotherapist to specify an appropriate joint range of
motion for the patient. The participant was instructed to sit on a
height-adjustable chair with the shank free on the leg holder,
with the hip and knee joints in 90° of flexion. His ankle-foot
complex was strapped into an ankle orthosis. The ankle orthosis
is rigidly connected with the foot plate.
The experimental data are plotted in Fig. 3 with satisfactory
trajectory tracking responses. The statistical results of the
trajectory tracking accuracy are summarized in TABLE I. For
ankle training in dorsiflexion and plantarflexion, the root mean
square deviation (RMSD) value is 0.0408 rad and the
(normalized root mean square deviation) NRMSD value is
8.16%. For ankle training in inversion/eversion and adduction/
abduction, the RMSD values are 0.0064 rad and 0.0714 rad,
respectively. It should be noted that the training in dorsiflexion
and plantarflexion was controlled while the training for the other
two DOFs was kept free. The trajectory deviation of training in
adduction/abduction may be caused by the foot abnormality.

The ankle robot was operated in a passive mode using the TABLE I. THE STATISTICAL TRAJECTORY TRACKING PERFORMANCE OF
joint space controller. The trajectory of ankle training is a sine THE ANKLE REHABILITATION ROBOT

wave along dorsiflexion and plantarflexion, with the frequency Motions Tracking Accuracy
being 0.02 Hz. The amplitude of the sine wave was initially set Dorsiflexion/Plantarflexion
RMSD (rad) 0.0408
at 0.1 rad, and then gradually increased until a feeling of joint NRMSD (%) 8.16
tightness. During the training, the subject was verbally RMSD (rad) 0.0064
Inversion/Eversion
NRMSD (%) NA
encouraged to relax his foot to minimize the effects by active
RMSD (rad) 0.0714
contributions. The training trajectories of inversion/eversion and Adduction/abduction
NRMSD (%) NA
adduction/abduction are set zero. The whole process lasted 15 RMSD: Root mean square deviation; NRMSD: Normalized root mean square deviation; RMSD and
NRMSD are defined in (5) and (6), where ¨ is the range of experimental values defined as the difference
minutes with 18 cycles. between the maximum and the minimum values in a data set. NA: Not applicable.

III. RESULTS
௡
One of the important functions of rehabilitation robots is to
guide the patient’s affected joint through certain position ‫ ܦܵܯ‬ൌ ඩ෍ሺ݉௜ െ ݁௜ ሻଶ Ȁ݊ (5)
trajectories. In this study, the position controller of the ankle ௜ୀଵ
robot was developed in joint space. Experimental results on the
participant are presented in Fig.3. In the first 100 seconds, the
training trajectory has an amplitude of 0.1 rad. Based on the ܴ‫ܦܵܯ‬
feeling of the patient, the range of motion was gradually ܴܰ‫ ܦܵܯ‬ൌ ൈ ͳͲͲΨ (6)
ο
increased until the patient felt tight at the ankle joint. During the
period of 100 to 200 seconds, the amplitude of the trajectory was
 The patient gave positive feedack in using this robot for
ankle stretching exercises, although some issues exist and may
have affected the rehabilitation efficacy. The biggest issue is the
fixation of the human foot during the training. When large torque
is applied to the human ankle, for example in extreme
dorsiflexion, the strap may become loose and the patient’s heel
will be lifted up. This could have made the actual ankle motion
different with the predefined trajectory due to relative movement
between the footplate and the human foot. This can be
considered as a limitation of this device when used for ankle
stretching.

0.3
Xd Xm

0.2
X Postion (rad)

0.1

-0.1

-0.2

0 100 200 300 400 500 600 700 800 900

Y Postion (rad)

0.2

-0.2 Yd Ym

0 100 200 300 400 500 600 700 800 900

0.2
Z Postion (rad)

-0.2 Zd Zm

0 100 200 300 400 500 600 700 800 900

Time (s)

Fig. 3. The trajectory tracking responses in task space during the robot-assisted ankle stretching. (,  and Z refer to ankle dorsiflexion/plantarflexion,
inversion/eversion, and adduciton/abduction, respectively. The subscript † and  represent desired and measured, respectively.)

IV. CONCLUSIONS AND FUTURE WORK
This study involves the use of an intrinsically compliant
rehabilitation robot for ankle stretching on patients with drop
foot. A stroke patient participated in the trial as a case study.
Results show this ankle rehabilitation robot can accurately and
reliably stretch the patient’s ankle joint to a specified position.
Preliminary findings using this ankle robot are promising for the
treatment of drop foot and support its clinical applications.
Future work will investigate the effectiveness of this ankle
robot for the treatment of drop foot on a large sample of patients.
The fixation of ankle joint during the training will be also solved
to allow more accurate ankle exercises following the predefined
trajectory.
ACKNOWLEDGMENT
This material was based on work supported by the University
of Auckland, Faculty of Engineering Research Development
Fund 3625057 (Physical Robot-Human Interaction for
Performance-Based Progressive Robot-Assisted Therapy) and
the China Sponsorship Council.
REFERENCES
[1] "Drop Foot," Wikipedia. Available:
https://en.wikipedia.org/wiki/Foot_drop
[2] J. D. Stewart, "Foot drop: Where, why and what to do?," Practical
Neurology, vol. 8, pp. 158-169, 2008.
[3] D. Mozaffafian et al., "Heart disease and stroke statistics—2015 Update,"
Circulation, vol. 131, pp. e29-e322, 2015.
[4] "Facts about stroke in New Zealand," Stroke Foundation of New Zealand.
Available: http://www.stroke.org.nz/stroke-facts-and-fallacies
[5] "Annual Report 2015," Stroke Foundation of New Zealand2015.
Available: http://www.stroke.org.nz/resources/2015%20Stroke%20AR-
web.pdf
[6] "New Zealand Spinal Cord Impairment Action Plan 2014-2019,"
Wellington of New Zealand, 2014. Available:
http://www.acc.co.nz/PRD_EXT_CSMP/groups/external_providers/doc
uments/papers_plans/wpc134157.pdf
[7] K. A. Shorter, G. F. Kogler, E. Loth, W. K. Durfee, and E. T. Hsiao-
Wecksler, "A portable powered ankle-foot orthosis for rehabilitation,"
Journal of Rehabilitation Research and Development, vol. 48, pp. 459-
472, 2011.
[8] J. Ward, T. Sugar, A. Boehler, J. Standeven, and J. R. Engsberg, "Stroke
Survivors' Gait Adaptations to a Powered Ankle-Foot Orthosis,"
Advanced Robotics, vol. 25, pp. 1879-1901, 2011.
[9] A. Faraji and M. R. Valajoozi, "Interactive foot orthosis (IFO) for people
with drop foot," in Applied Mechanics and Materials vol. 464, ed, 2014,
pp. 129-134.
[10] A. I. R. Kottink, L. J. M. Oostendorp, J. H. Buurke, A. V. Nene, H. J.
Hermens, and M. J. Ijzerman, "The Orthotic Effect of Functional
Electrical Stimulation on the Improvement of Walking in Stroke Patients
with a Dropped Foot: A Systematic Review," Artificial Organs, vol. 28,
pp. 577-586, 2004.
[11] P. H. Peckham and J. S. Knutson, "Functional Electrical Stimulation for
Neuromuscular Applications," Annual Review of Biomedical
Engineering, vol. 7, pp. 327-360, 2005.
[12] G. M. Lyons, T. Sinkjaer, J. H. Burridge, and D. J. Wilcox, "A review of
portable FES-based neural orthoses for the correction of drop foot," Ieee
 Transactions On Neural Systems And Rehabilitation Engineering, vol. 10,
pp. 260-279, 2002.
[13] J. S. Yeap, R. Birch, and D. Singh, "Long-term results of tibialis posterior
tendon transfer for drop-foot," International Orthopaedics, vol. 25, pp.
114-118, 2001.
[14] V. Krishnamoorthy et al., "Gait training after stroke: A pilot study
combining a gravity-balanced orthosis, functional lectrical stimulation,
and visual feedback," Journal of Neurologic Physical Therapy, vol. 32,
pp. 192-202, 2008.
[15] N. Romansky, K. Scollon-Grieve, and J. G. McGinness, "Current
Concepts In Diagnosing And Treating Drop Foot," vol. 25, 2012.
[16] N. Takeuchi and S.-I. Izumi, "Rehabilitation with poststroke motor
recovery: a review with a focus on neural plasticity," Stroke Research and
Treatment, vol. 2013, p. 128641, 2013.
[17] P. K. Jamwal, "Design analysis and control of wearable ankle
rehabilitation robot," Doctor of Philosophy PHD thesis, Mechanical
Engineering Department, The University of Auckland, New Zealand,
2011.
[18] A. Roy et al., "Robot-aided neurorehabilitation: a novel robot for ankle
rehabilitation," IEEE Transactions on Robotics, vol. 25, pp. 569-582,
2009.
[19] Y. L. Park et al., "Design and control of a bio-inspired soft wearable
robotic device for ankle-foot rehabilitation," Bioinspiration and
Biomimetics, vol. 9, p. 016007, 2014.
[20] L.-Q. Zhang et al., "Intelligent stretching of ankle joints with contracture
spasticity," IEEE Transactions on Neural Systems and Rehabilitation
Engineering, vol. 10, pp. 149-157, 2002.
[21] P. K. Jamwal, S. Q. Xie, S. Hussain, and J. G. Parsons, " An adaptive
wearable parallel robot for ankle injury treatments," IEEE Transactions
on Mechatronics, vol. 19, pp. 64-75, 2014.
[22] Y. H. Tsoi, S. Q. Xie, and A. E. Graham, "Design, modeling and control
of an ankle rehabilitation robot," Studies in Computational Intelligence,
vol. 177, pp. 377-399, 2009.
[23] J. A. Saglia, N. G. Tsagarakis, J. S. Dai, and D. G. Caldwell, "A high-
performance redundantly actuated parallel mechanism for ankle
rehabilitation," The International Journal of Robotics Research, vol. 28,
pp. 1216-1227, 2009.
[24] M. Girone, G. Burdea, M. Bouzit, V. Popescu, and J. E. Deutsch, "A
stewart platform-based system for ankle telerehabilitation," Autonomous
Robots, vol. 10, pp. 203-212, 2001.
[25] M. Zhang, T. C. Davies, and S. Xie, "Effectiveness of robot-assisted
therapy on ankle rehabilitation - a systematic review," Journal of
NeuroEngineering and Rehabilitation, vol. 10, p. 30, 2013.
[26] J. Yoon, J. Ryu, and K.-B. Lim, "Reconfigurable ankle rehabilitation
robot for various exercises," Journal of Robotic Systems, vol. 22, pp. 15-
33, 2006.
[27] L. Gengqian, G. Jinlian, Y. Hong, Z. Xiaojun, and L. Guangda, "Design
and Kinematics Simulation of Parallel Robots for Ankle Rehabilitation,"
in Mechatronics and Automation, Proceedings of the 2006 IEEE
International Conference on, 2006, pp. 1109-1113.
[28] C. E. Syrseloudis and I. Z. Emiris, "A parallel robot for ankle
rehabilitation-evaluation and its design specifications," in 8th IEEE
International Conference on BioInformatics and BioEngineering, 2008,
pp. 1-6.
[29] M. Hassan and A. Khajepour, "Analysis of bounded cable tensions in
cable-actuated parallel manipulators," IEEE Transactions on Robotics,
vol. 27, pp. 891-900, 2011.
Muhammad Ihsan Hanif
081811733035
Rehabilitation Device

Alat ini menggunakan aktuator berupa Fluidic Muscle (FESTO DMSP-20-400N) yang berbasis
penumatic actuator. Terdapat 4 aktuator serupa yang bisa digunakan untuk melakukan latihan
dengan 3 DOF (plantarflexion/dorsiflexion, eversion/inversion, adduction/abduction). Terdapat
juga 4 single-axis load cell dan 1 six-axis load cell untuk mengetahui besarnya gaya dan torsi
yang dihasilkan yang dapat digunakan untuk analisis.
Terdapat aktuator yang tetap (fixed) dan ada juga yang dapat bergerak. Setiap aktuator diberi
regulator tekanan untuk mengatur tekanan pada aktuator sehingga dapat memanjang atau
memndek. Selain itu, terdapat juga 3 rotary encoder untuk mengetahui letak 3 dimensi dari
telapak kaki pasien sehingga bisa melakukan latihan dengan presisi. Fisioterapis akan
memberikan informasi mengenai lintasan yang dibutuhkan pasien saat latihan yang kemudian
nanti akan dikonversi kedalam tekanan yang dibutuhkan aktuator untuk memanjang atau
memendek menggunakan metode inverse kinematics.
Perbedaan dari Assistive device sebelumnya adalah pada rehabilitation device tidak terdapat
sensor yang memberikan input kepada aktuator secara terus menerus, melainkan input sudah
ditetapkan di awal (predefined) oleh fisioterapi. Hal itu disebabkan karena rehabilitasi
membutuhkan pergerakan yang berulang berulang dengan kondisi yang sama persis sehingga
bisa membuat otot lebih terbiasa. Selain itu, dari desainnya juga terdapat perbedaan yang
signifikan, assistive device bisa portable sedangkan rehabilitation device tidak. Rehab device
membutuhkan hasil yang maksimal dalam proses latihan sehingga perlu kondisi yang ketat
sehingga banyak komponen yang harus fixed, sedangkan pada assist device membutuhkan
sesuatu yang ringan dan bisa dibawa kemana mana yang cukup untuk membuat orang berjalan
dengan normal.