Available via license: CC BY 4.0
Content may be subject to copyright.
1
Underactuated Base-to-Distal Hand Exoskeleton for
Adaptive Grasping Assistance
Wenyuan Chen, Guangyong Li, Member, IEEE, Mingwei Li, Xin Li, Wenxue Wang, Member, IEEE,
Xiujuan Xue, Xingang zhao, and Lianqing Liu, Member, IEEE,
Abstract—Hand exoskeletons are wearable devices that can
provide outer kinematic coupling with human hands and thus
assist movement of human fingers. However, conventional rigid
hand exoskeletons are characterized by their bulky and complex
structures, which are often incompatible with human finger
joints and restrict finger’s natural motion. This paper reports
an underactuated base-to-distal hand exoskeleton that provides
adaptive grasping assistance. An underactuated 8-bar base-to-
distal linkage driven by a cable is used to flex and extend fingers
and it applies force only to the distal phalanges of fingers, which
not only makes the hand exoskeleton adapt to different sizes of
fingers but also allows all phalanges to naturally accommodate
the geometry of the objects to be grasped. The kinematic model of
the 8-bar linkage is derived in order to generate desired hand ges-
tures. A five-finger hand exoskeleton with active flexion/extension
(F/E) for all fingers and active abduction/adduction (Ab/Ad) for
the thumb is assembled and then tested on a healthy subject
and a stroke survivor. Experimental results show that the hand
exoskeleton can generate sufficient fingertip force for regular
tasks. The hand exoskeleton enables the healthy participant and
the stroke survivor to achieve 90% and 52% of their passive
range of F/E motions respectively. In addition, the stroke survivor
can accomplish various training tasks, such as grasping, pinching
and writing, with the assistance of the hand exoskeleton. These
results demonstrate that the underactuated base-to-distal hand
exoskeleton can be an effective device for rehabilitation training
or daily-life assistance for patients with a hemiparetic hand.
Index Terms—Hand exoskeleton, grasp assistance, base-to-
distal exoskeleton, rehabilitation robot, underactuated robot.
I. INTRODUCTION
Stroke is a serious neurological injury that damages the
brain, leading to disability or even causing death [1]. About
12.2 million incident strokes are diagnosed each year, resulting
in more than 100 million prevalent stroke survivors worldwide
This work is supported by the National Natural Science Foundation of
China (Grant Nos. 61925307, U1908215, 62003338, 61933008, 61821005),
CAS Project for Young Scientists in Basic Research (Grant No. YSBR-041)
and Liaoning Revitalizaiton Talents Program (Grant No. XLYC2002014).
W. Chen is with the State Key Laboratory of Robotics, Shenyang Institute of
Automation, Chinese Academy of Sciences (CAS), Shenyang 110016, China,
Institutes for Robotics and Intelligent Manufacturing, Chinese Academy
of Sciences, Shenyang, 110016, China and the University of the Chinese
Academy of Sciences, Beijing 100049, China (e-mail: chenwenyuan@sia.cn).
G. Li is with Department of Electrical and Computer Engineering, Swanson
School of Engineering, University of Pittsburgh (e-mail: gul6@pitt.edu).
M. Li and X. Li is with Suzhou Xianna Precision Instr. Co. LTD, Jiangsu,
China (e-mail: 874186336@qq.com; lx13331677467@163.com.
W. Wang, X. Zhao and L. Liu are with the State Key Laboratory of
Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences
(CAS), Shenyang 110016, China (e-mail: wangwenxue@sia.cn; zhaoxin-
gang@sia.cn; lqliu@sia.cn).
X. Xue is with Rehabilitation Center for the Disabled, Shenyang, 110015,
China. (e-mail: xxj190003@163.com)
* L. Liu and G. Li are the corresponding authors.
[2]. Many stroke survivors lose muscle control and sensation
in their upper extremities either temporarily or permanently.
Hand function accounts for 90% of the upper limb function
[3], [4], and thus its loss severely affects the activities of
daily living (ADLs) of post-stroke survivors and drastically
diminishes their life quality. Although some survivors can
regain their arm function through repetitive motion training,
it is difficult to recover their hand function since the hand is
furthest from the midline of the body and has the longest
neural-muscle control pathways [5]. Therefore, continuous
active rehabilitation training through repetitive movements is
needed for stroke survivors in order to further recover their
hand functions. Many studies have suggested that robotic as-
sistance and rehabilitation devices can lead to motor recovery
in upper limbs, which is at least comparable, if not superior,
to rehabilitation training by occupational therapists [6], [7].
A variety of hand exoskeleton devices have been developed
for hand motion rehabilitation or assistance in ADLs [8]–
[11]. Currently, two kinds of hand exoskeletons have been
intensively investigated: soft gloves driven by pneumatic ac-
tuators or tendons (cables) and rigid link-based exoskeletons
driven by gears or cables. Although soft gloves with pneumatic
actuators have been commercialized as passive training devices
already, they are usually not capable of generating sufficient
gripping force for patients with paralyzed hands to grasp and
move heavier objects [12]–[15]. In addition, the pneumatically
driven soft gloves in general require a bulky pump and thus
are not easy to carry. Soft tendon-actuated gloves have drawn
much attention recently for their light weight and compact size
[16]–[21]. Our previous studies [22], [23] have demonstrated
that stroke survivors can benefit from soft tendon-actuated
gloves in rehabilitation training and ADLs. However, soft
tendon-actuated gloves require a large number of tendons on
both sides of hand in order to actuate thumb and fingers
motions, which causes a challenge for patients to don and
doff the soft gloves by themselves. Especially the tendons on
palmar side result in difficulty in grasping objects, and those
on dorsal side make patients feel uncomfortable with pressing
force along the finger during extension. Additionally, our test
revealed that the flexion range of fingers under the assistance
of soft tendon-actuated gloves is much lower than that of
healthy fingers, which is not conducive to the recovery of
patients’ hand functions. Rigid hand exoskeletons can produce
much stronger grip strength, but they are usually heavy and
bulky and have to be designed customizedly because of the
differences in hand size. In addition, most rigid hand exoskele-
tons lack compliance with human natural motions because of
2
joint misalignment between exoskeletons and human fingers
[24]. For this reason, some rigid hand exoskeletons adopt some
innovative mechanisms that can self-align robotic joints with
human finger joints to prevent user’s fingers from injury [25],
[26]. However, these mechanisms make hand exoskeletons
even heavier and bulkier.
Experimental results suggest that distal training is benefi-
cial for the recovery of whole upper limb [27]. Therefore,
rehabilitation training with base-to-distal hand exoskeletons
could potentially expedite the recovery of hand functions. In
addition, the base-to-distal hand exoskeletons do not require
joint alignment between the fingers and the exoskeleton. A few
studies on rigid link-based hand exoskeletons have reported to
adopt the base-to-distal mechanism in their design [28]–[31].
Unfortunately, most of them are still in their pilot stage with
only F/E motions for two or three fingers [28], [29], [31], or
for five fingers with one actuator [30]. Furthermore, none of
these designs have yet considered the thumb Ab/Ad motions,
which are critically important for hand functions.
In this paper, we report a light and compact base-to-distal
hand exoskeleton for assisting both finger flexion/extension
and thumb abduction/adduction based on an underactuated 8-
bar linkage mechanism. The hand exoskeleton is mounted on
the dorsal side of hand and applies force only to the distal
phalanx of fingers. Users can easily don and doff the hand
exoskeleton by themselves. The hand exoskeleton provides
kinematic adaptability not only to different sizes and stiffness
of hands but also to different geometries of objects to be
grasped. Therefore, it can fit all hand sizes and accommodate
objects with different geometries. Similar to soft tendon-
actuated gloves, the hand exoskeleton is driven by remote actu-
ators through cables and tubes. We evaluate the effectiveness
of the base-to-distal hand exoskeleton on a healthy subject
and a stroke survivor. The results demonstrate that the hand
exoskeleton is able to adapt to different sizes of objects during
grasping operation and can generate sufficient fingertip force
for regular tasks. The hand exoskeleton significantly enhances
the stroke survivor’s ability to perform some typical ADLs,
such as grasping, pinching and writing.
The rest of the paper is organized as follows. Section II
introduces the detailed design of the hand exoskeleton sys-
tem; Section III describes the kinematic analysis and control
method of the hand exoskeleton; Section IV provides the
experimental results and further discussion. The conclusion
is finally presented in Sections V.
II. SYSTEM DESIGN
The entire hand exoskeleton system, as shown in Fig. 1,
is composed of three modules: 1) a wearable base-to-distal
hand exoskeleton that assists the movements of fingers and
thumb; 2) a remote actuation system (RAS) that drives the
hand exoskeleton through cables, which can be tied around
the waist; 3) a microcomputer control system that controls the
RAS through Bluetooth. The base-to-distal hand exoskeleton
alone weighs approximately 235g, and the entire system 1.4kg
(including electronics and batteries).
The design of the hand exoskeleton system is described
below with regard to the exoskeleton finger mechanism, the ex-
Remote
actuation system Bluetooth
Microcomputer
control system
Base-to-distal
hand exoskeleton
Fig. 1. The entire hand exoskeleton system is composed of three modules: a
wearable base-to-distal hand exoskeleton, a remote actuation system (RAS),
and a microcomputer control system.
oskeleton thumb mechanism, and the whole hand exoskeleton
mechanism, respectively, as followed by the remote actuation
system and the microcomputer control system.
A. Exoskeleton Finger Mechanism Design
Hand exoskeletons should be able to assist human fingers
to move in the full workspace, and be compatible with
their natural motions. Except for the thumb, every finger
has four joints: Distal Interphalangeal (DIP) joint, Proximal
Interphalangeal (PIP) joint, Metacarpophalangeal (MCP) joint,
and Carpometacarpal (CMC) Joint, each with one degree of
freedom (DOF) in flexion/extension (F/E) direction. However,
for these fingers, the CMC joints have little F/E motions and
thus are usually not considered in the design of hand ex-
oskeleton. In most studies, a common design for rigid linkage-
based hand exoskeletons is to build a linkage mechanism that
couples the F/E motion of DIP, PIP, and MCP joints so as to
reduce the number of actuators. However, such mechanism in-
evitably disrupts natural motions of human fingers may cause
injury to users because it is hard to precisely measure and
control the coupling relationship for users with various finger
sizes, especially when interacting with objects with different
geometries. Additionally, the range of F/E motion in human
fingers is large, with normal ranges of 120◦, 100◦, and 80◦for
MCP, PIP, and DIP, respectively [32]–[34]. Unfortunately,
most finger mechanisms of conventional hand exoskeletons
are not capable of achieving such large range of motion
(ROM) [35], [36]. Moreover, the MCP joint has an additional
abduction/adduction (Ab/Ad) DOF, which contributes to the
stability and flexibility of grasping but is often neglected in
the design of hand exoskeletons.
To overcome the limitations of traditional design mechanism
for rigid linkage-based hand exoskeletons, we developed an
underactuated 8-bar base-to-distal linkage as shown in Fig.
2(a) and Fig. 3(a). The linkage has two bars crossing each
other with its end attached to the fingertip while its base
mounted on a back plate on the dorsal side of hand. The un-
deractuated 8-bar linkage mechanism is able to generate finger
3
F/E motion through a cable and a torsion spring. The finger
flexion is achieved via pulling a cable by a remote motor,
and the finger extension by a torsion spring in combination
with releasing the cable by the motor. Specifically, the finger
flexion is achieved as the cable is pulled by a remote motor and
meanwhile the torsion spring is compressed to restore potential
energy, and the finger extension is performed as the cable is
released and the torsion spring drives the linkage back to its
rest position. The 8-bar base-to-distal linkage, with properly
selected link lengths, allows the hand exoskeleton to fit various
finger sizes to some extent and is able to easily achieve the
entire range of normal finger F/E motion.
(c)(b)
(a)
Finger
Ad/Ab Joint
8-bar linkage
mechanism
Cable
Fixed point
Spring
Abbuction
Adbuction
Joint
Link-1
Cable
Joint with
a spring Tub e
Pulley
Motor
Pulley
Link-1
Fig. 2. The schematic diagrams of the proposed mechanisms of the hand
exoskeleton. (a) The underactuated base-to-distal finger mechanism for F/E
motions of fingers and thumb. (b) Revolute joints for Ab/Ad motions of
fingers. (c) The thumb mechanism for thumb Ab/Ad motions.
(a) (c)(b)
8-bar linkage mechanism
Dorsal side
Spring
Abbuction
Adbuction
Joint
Finger
Ad/Ab joint
Cables and tubes
8-bar linkage mechanism
Fig. 3. The proposed hand exoskeleton is worn by the participant. (a) The
underactuated base-to-distal mechanism for F/E motions of fingers and thumb.
(b) Revolute joints for Ab/Ad motions of fingers. (c) The thumb mechanism
for thumb Ab/Ad motions.
The planar DOF of a linkage in combination with a finger
can be calculated by Grubler’s formula [31] as below:
DOF =3(n−1) −2f1−f2(1)
where nis the total number of the links and finger pha-
langes, f1the revolute joints and f2the higher-order joints,
respectively. In this study, the 8-bar linkage combined with
a finger has n=10links, f1=12revolute joints and
f2=0higher-order joints, and therefore its planar DOF is
3. Since the 8-bar linkage is driven by only one actuator, it is
highly underactuated. The underactuated mechanism enables
indefinite planar motion with respect to all the links and finger
phalanges through the cable. During the finger flexion and
extension assisted by the underactuated 8-bar linkage mecha-
nism, the force is transmitted to the fingertip only through the
links and does not limit the finger movements in the planar
motions, which means that the linkage will not force the finger
to move along a definite trajectory but bring flexibility in
finger motion within a narrow band constrained by the finger
itself. Especially, when the finger interacts an object during
grasping operation, the finger phalanxes attach to the object
causing constraints on finger joints, and thus the equivalent
planar DOF of the linkage in combination with the finer
drops to one. Therefore, the underactuated linkage mechanism
provides grasping adaptivity to reconfigurable hand stiffness
and posture so as to accommodate distinct geometries of
grasped objects.
In addition to finger F/E motion, finger Ab/Ad motion is
also essential for stable and comfortable grasping. A revolute
joint is used to connect the 8-bar linkage mechanism through
its first link (Link-1) to the back plate as shown in Fig. 2(b)
and 3(b), and therefore the linkage mechanism is not just fixed
but allowed to rotate around the revolute joint by the inherent
finger Ab/Ad motions. In another word, the 8-bar linkage
mechanism will not prevent the finger from its inherent Ab/Ad
motions. As a result, the exoskeleton finger mechanism can
assist patients to perform both active F/E motion and passive
Ab/Ad motion of fingers.
B. Exoskeleton Thumb Mechanism Design
The human thumb has three joints: CMC joint, MCP joint,
and Interphalangeal (IP) joint. Similar to the other fingers, the
thumb MCP and IP joints have one F/E DOF. However, it is
challenging to accurately describe the motion of the thumb
CMC joint for its complex anatomy and sophisticated kine-
matic nature. To simplify the exoskeleton thumb mechanism
and enable it to perform the principle motions of a thumb, the
thumb CMC joint is considered to have one F/E DOF and one
Ab/Ad DOF in this study.
We developed a thumb mechanism, consisting of two parts
that enable active thumb F/E and Ab/Ad motions respectively.
The active thumb F/E mechanism is identical to the 8-bar
linkage mechanism for other fingers described in Section II-A.
And, similar to the exoskeleton finger mechanism, the 8-bar
linkage mechanism is installed on a side plate by a revolute
joint at the extended end of the first link (Link-1) as shown in
Fig. 2(c) and 3(c). The active thumb Ab/Ad motion is realized
through a cable and a linear spring that are connected to Link-
1, and allow the 8-bar linkage mechanism to rotate around
the revolute joint as shown in Fig. 2(c) and 3(c). Specifically,
the thumb abduction is performed as the cable is pulled and
4
the spring is stretched, and the thumb adduction is executed
as the cable is released and the spring returns back by the
restoring force. By the active thumb Ab/Ad mechanism, the
angle range of thumb Ab/Ad motions can be enlarged to ensure
the requirement for hand operation that a thumb should have
much larger range of Ab/Ad motion than the other fingers
[37], [38]
C. Whole Hand Exoskeleton
A prototype of a whole hand exoskeleton was assembled
with the exoskeleton mechanisms of fingers and thumb de-
scribed in Section II-A and II-B, respectively. A back plate
and a side plate, which are designed according to the shape
and size of user’s hand and manufactured by 3D printing, are
used for the assembly of the exoskeleton linkage mechanisms.
A soft fabric glove is sewn to the hand plates in order to
ensure comfortable contact with hand skin and also provide
stable mounting between linkage ends and fingertips during
flexion. Five 8-bar linkage mechanisms corresponding to the
thumb and fingers are mounted on the hand dorsal plate
through revolute joints. All links of the 8-bar linkages are
made of aluminum, which makes the exoskeleton lightweight
at just 235g. The hand exoskeleton is fastened by Velcro straps
and can be donned and doffed easily with another hand. In
addition, the fishing lines with high output force are chosen as
the pulling cable. Each 8-bar linkage is actuated by one cable
to perform the finger/thumb F/E motions, an additional cable
is used to actuate the thumb Ab/Ad motions, and therefore 6
cables in total are used in the actuation.
D. Remote Actuation System and Microcomputer Control Sys-
tem
A remote actuation system (RAS) that drives the hand
exoskeleton consists of the following items: six micro servo
motors (LDX-227, Hiwonder, China), six pulleys, a custom-
made motor driver board (LSC-6, Hiwonder, China), and a
20000mAh Li-battery. Each motor can output a maximum
torque of 15 kg·cm and its rotation angle ranging 0-270◦can be
accurately controlled by the driver board so as to actuate the 8-
bar linkage. All electronic components of RAS are powered by
the Li-battery that enables the RAS to work for approximately
5 hours. The RAS is compact and can be conveniently worn
around the waist of users.
A microcomputer control system, which is connected to the
RAS through Bluetooth and thus can be placed anywhere,
provides a touchscreen Graphical User Interface (GUI) for
users to control the RAS to drive the hand exoskeleton.
Through the microcomputer control system, the finger/thumb
F/E motions and the thumb Ab/Ad motions can be controlled
independently, and specific training paradigms with coordi-
nated finger motions can also be executed. Before executing
the training paradigms, the motor rotation angles for hand
training movements are calibrated in advance. And the hand
training speed and time can also be set by users.
III. KINEMATIC ANALYSIS AND SYSTEM CONTROL
The control mechanism of the hand exoskeleton, imple-
mented in the microcomputer control system, is shown in Fig.
4. The control of each finger motion is based on the kinemat-
ical model of the underactuated 8-bar linkage mechanism. In
this section, the index finger control mechanism is chosen as an
example for illustration purposes, and the control mechanisms
for other fingers are the same except that the parameter values
are different. In this section, the kinematic analysis of the
exoskeleton finger/thumb mechanism for F/E motion is firstly
provided in detail, and then the control mechanism of the hand
exoskeleton is described followed by the verification of the
control mechanism through simulation. It is worth noting that
the kinematics of the exoskeleton thumb Ab/Ad motion is not
analyzed since it is simple.
Fig. 4. The control mechanism of the hand exoskeleton.
A. Kinematics of the Exoskeleton Finger Mechanism
For the convenience of analysis, the exoskeleton finger
mechanism is divided into two parts: a 4-bar linkage chain as
shown in Fig. 5(a) and a 5-bar linkage chain as shown in Fig.
5(b). Correspondingly, the link lengths li(i=1,2,...,10), the
link rotation angles qj(j=1,2,...,7), the fingertip position
(V,H)and the finger flexion angles θk(k=1,2,3) are defined
in Fig. 5. Because the exoskeleton finger mechanism is an
underactuated system, a single control input with respect to q1
cannot be uniquely determined from two outputs with respect
to the finger position (V,H). In the study, a backward analysis
is performed so as to determine the control input q1given a
pair of output values (V,H)such that
q1=f(V, H)(2)
We hypothesize that the phalanx lengths dm(m=1,2,3)
and the flexion angles of a finger θk(k=1,2,3) are known
when grasping an object, then the finger position (V,H)can
be calculated by the following equation:
V=d1c1+d2c12 +xTc123 +yTs123 −x0
H=d1s1+d2s12 +xTs123 −yTc123 +y0(3)
5
where (x0,y
0)and (xT,y
T)are the coordinates of MCP and
DIP joints respectively and,
c1= cos (θ1)
c12 = cos (θ1+θ2)
c123 = cos (θ1+θ2+θ3)
s1=sin(θ1)
s12 =sin(θ1+θ2)
s123 =sin(θ1+θ2+θ3)
Apparently, the output values (V,H)are related to the input
q1through a middle value q4. Their relations can be obtained
by analyzing the 4-bar linkage chain and the 5-bar linkage
chain separately.
(a) (b)
l1
l2
l3
l4q1
q2
q4
l4
l5
l6
l7
l8
l9
l10
q5
q6
q7
φ
V
H
θ1
θ2
θ3
MCP
PIP
DIP
x0
y0
yT
xT
4-bar linkage chain 5-bar linkage chain
Fig. 5. The schematic of the 8-bar base-to-distal linkage mechanism and
parameter definitions used in the kinematic model. The mechanism is divided
into two parts: (a) a 4-bar linkage chain and (b) a 5-bar linkage chain.
The close-loop kinematic equation for the 4-bar linkage
chain as shown in Fig. 5(a) is given by
l2eq1i+l3e−q2i+l4e(π+q4)i=l1(4)
From Eq. (4),the relationship between q1and q4can be written
as below:
q1=g(q4)=arcosC
√A+B+β(5)
where
A=−2l2l4cos q4
B=−2l2l4sin q4
C=−l12−l22+l32−l42−2l1l4cos q4
β=arctan
B
A
The close-loop kinematic equation for the 5-bar linkage
chain as shown in Fig. 5(b) is given by
(l4+l5)e(q4+ϕ)i+l6eq5i+l4e−q7i=T
(l4+l5+l6)e(q4+ϕ)i+l6e−q6i+l4e−q7i=T(6)
where
T=H2+V2
Based on Eq. (6), the relationship between q4and (V,H)
q4=h(V, H)(7)
can be solved from the following formula:
F2−2EF cos (q4+ϕ)+E2−sin (q4+ϕ)2=0 (8)
where
A1=(l4+l5)2+(l7+l8)2+T2−l2
9
A2=(l4+l5+l6)2+l2
8+T2−l102
B1=−A1
2+(l4+l5)·T·cos (q4+ϕ)
l7+l8
B2=−A2
2+(l4+l5+l6)·T·cos (q4+ϕ)
l8
E=B2−B1
l6
F=(l4+l5)·B2−(l4+l5+l6)·B1
l6·T
As a result, the control value of finger mechanism q1can
be calculated through Eqs. (3), (5) and (7). Then, the change
of the cable length ΔLcan be obtained from the following
equation:
ΔL=L0−l2
1+l2
2−2l1l2cos q1(9)
where L0is the value of the cable length in the case of full
extension of the finger. Because the pulley and the actuation
motor are co-axial, the rotation angle of the motor θmotor can
be calculated by the following equation:
θmotor =ΔL
r(10)
where ris the radius of the pulley.
B. Control Mechanism of the Hand Exoskeleton
The control mechanism for the hand exoskeleton is shown
in Fig. 4. The microcomputer control system provides a touch-
screen GUI for users to choose a specific training paradigm,
calibrate the final finger angles of training paradigms, and set
the training speed and time. The trajectory planner constructs
the F/E motion trajectories in terms of angles for the fingers
and the thumb according to the selected training paradigm,
and the corresponding rotation angles of the motors θmotor are
calculated by Eq. (10). Then, θmotor is converted into the po-
sition command of the servo motors, the motor controller will
control servo motors to follow the position profiles, and the
corresponding motions of the hand exoskeleton are generated.
Once the measured torque is beyond the predefined maximum
torque, the controller stops the servo motors immediately to
prevent the fingers of users from injury.
C. Simulation of Hand Motions Assisted by the Linkage
Hand motions assisted by the base-to-distal linkage mech-
anisms were simulated with the Simscape simulator of MAT-
LAB software (Mathworks) as shown in Fig. 6 and the
Supplementary Video. The following three scenarios, 1) flex-
ing/extending fingers, 2) grasping, and 3) pinching, were
simulated to confirm that the base-to-distal linkage mechanism
allows all phalanges to naturally move and to accommodate
the geometry of the object to be grasped. For convenience,
the ranges of all finger joints were set to 0–90◦and the base-
to-distal linkage mechanisms are actuated by the active joint
rotation angle q1.
6
Fig. 6. The simulation of the hand motions assisted by the base-to-distal
linkage mechanisms. (a) Full Extension; (b) Grasping a large cylinder; (c)
Grasping a small cylinder; (d) Pinching a small cylinder.
1) Flexing/Extending fingers: The finger is flexed and
extended by applying a torque to the active joint q1through
the cable counterclockwise and the torsion spring clockwise,
respectively. The underactuated base-to-distal linkage mech-
anism is capable of adapting to the finger joint system as a
whole and its equivalent stiffness, and thus allows the finger to
naturally flex and extend. Full finger extension and flexion can
both be achieved by the assistance of the linkage mechanism
as shown in Fig. 6(a) and the Supplementary Video.
2) Grasping: As shown in Fig. 6(b) and (c), two cylin-
ders with different sizes (large and small) are grasped by
the underactuated base-to-distal linkage mechanism. A stable
grasping posture needs to be determined so as to adapt to the
geometry of the cylinder. The cross section of the cylinder
is a circle centered at (xc,yc) with a diameter of r. The
length and the diameter of each finger phalanx are denoted
as dm(m=1,2,3) and rpm(m=1,2,3), respectively. The
line segments (x1,s2,x2and s3) as shown in Fig. 6(b) and
(c) can be calculated as follows:
x1=(xc2+yc2)−(r+rp1)2
s2=(r+rp1)2+(d1−x1)2
x2=(r+rp1)2+(d1−x1)2−(r+rp2)2
s3=(r+rp2)2+(d2−x2)2
As a result, the finger posture can be determined from the
following equations:
θ1=π
2−arctan yc
xc−arcsin r+rp1
x2
c+y2
c
θ2=π−arcsin r+rp2
s2−arctan r+rp1
d1−x1
θ3=π−arcsin r+rp3
s3−arctan r+rp2
d2−x2
As shown in Fig. 6(b) and (c) and the Supplementary Video,
the finger can adaptively grasp both large and small cylinders
with the assistance of the base-to-distal linkage mechanism.
3) Pinching: As shown in Fig. 6(d), a cylinder is used as
an instance to be pinched. The cross section of the cylinder
is a circle centered at (xc,yc) with a diameter of r. The
length and the diameter of each finger phalanx are denoted
as dm(m=1,2,3) and rpm(m=1,2,3), respectively. Then,
the coordinates of the contact point Pof the distal finger
phalanx on the cylinder can be determined by the position
of the cylinder as follows:
xP=xc+(r+rp3)·sin (θ1+θ2+θ3)
yP=yc−(r+rp3)·cos (θ1+θ2+θ3)(11)
The coordinates of Pcan also be calculated with the posture
of the finger as follows:
xP=d1cos θ1+d2cos (θ1+θ2)+x3cos (θ1+θ2+θ3)
yP=d1sin θ1+d2sin (θ1+θ2)+x3sin (θ1+θ2+θ3)
(12)
In addition, the following relation between θ3and θ2has
been reported in [39], [40]:
θ2=1.5·θ3(13)
The DOF of the finger mechanism decreases to 1 because
of constraints (11), (12) and (13). Therefore, the finger posture
for object pinching can be determined by one active actuation,
i.e., the active joint q1of the base-to-distal linkage mechanism.
As shown in Fig. 6(d) and the Supplementary Video, the finger
can adaptively pinch the small cylinder with the assistance of
the base-to-distal linkage mechanism.
IV. RESULTS AND DISCUSSIONS
The proposed hand exoskeleton is experimentally tested and
evaluated on a healthy participant and a stroke survivor. We
first test the mechanical properties (fingertip force level and
finger trajectories) of the finger mechanism to evaluate its
performance on grasping assistance. Furthermore, in order to
assess the effectiveness of the whole hand exoskeleton system
for rehabilitation training and daily-life assistance, the stroke
survivor is asked to grasp some common objects in daily life
with the assistance of the hand exoskeleton. All experiments
were conducted under the supervision and guidance of doctors
from the Rehabilitation Center for the Disabled of Liao Ning
province. The information of participants is summarized in
TABLE I.
7
TABLE I
INFORMATION OF THE PARTICIPANTS
Participant Healthy
Participant Stroke Survivor
Gender M M
Age 28 49
Affected (stroke survivor) or
Dominant (healthy
participant) side
Right Right
Duration of Stroke (years) – 8
Stroke Type – Ischemic Stroke
*FMA-UE – 16/66
*Fugl-Meyer Assessment Upper Extermity (FMA-UE)
A. Evaluation of Fingertip Force Level
In the first experiment, the index fingertip force of the stroke
survivor was measured to evaluate the force level of grasping
assistance with the 8-bar base-to-distal linkage mechanisms.
As shown in Fig. 7(a), the force sensor is placed on a stack
of paper whose overall thickness ΔLcan be conveniently
adjusted. The stroke survivor was asked to pinch the force
sensor and the paper stack with the assistance of the hand
exoskeleton and the fingertip force was recorded by the force
sensor. And the maximum force was acquired when the motor
stopped rotating due to the resistance. In this experiment, the
overall thickness ΔLvaries from 20mm to 80mm.
Finger mechanism
△L
(a) (b)
Force
sensor
Fig. 7. Evaluation of fingertip force Level. (a) The hand exoskeleton assists
the fingers of the stroke survivor to pinch the force sensor and a stack of
paper. (b) Experimental results of the pinching force of the fingertip.
The experimental results, as shown in Fig. 7(b), exhibit a
monotonic relationship between the maximum pinching force
assisted by the hand exoskeleton and the thickness of the
object to be pinched, and the maximum pinching force grows
as the thickness of the object increases. In this study, the
maximum pinching force of the stroke survivor ranges from
9.4N to 10.8N with the assistance of the hand exoskeleton,
indicating that the hand exoskeleton can provide a stable
grasping force and meets the requirement of rehabilitation and
grasping assistance.
B. Measurements of Finger Trajectories
The trajectories of finger flexion and extension motions are
measured by an optical capture system (OptiTrack, LEYARO,
China) which can track the locations of finger joints and some
joints of the 8-bar linkage mechanism by point markers. The
point markers are attached to 4 finger joints (J1,J2,J3and
EJ) and 4 linkage joints (P1to P4) as shown in Fig. 8.
J0
J1
J2
EJ
P1
P4
P2
P3
Fig. 8. The measurement of finger trajectories. The point markers are attached
to the finger joints and some of the joints of the 8-bar linkage mechanism to
track their locations as shown by labels.
In the experiment, the healthy participant and the stroke
survivor were asked to keep their arms and hands hanging and
immobile, and their index fingers performed linkage-assisted
flexion and extension motions three times. To compare the
finger trajectories of the healthy participant and the stroke
survivor, the trajectories of point markers were recorded during
the motions. Then, the rotational angles of three finger joints
(MCP, PIP and DIP) and the active joint (q1) of the linkage
mechanism were calculated accordingly, which is described in
Appendix. The experimental results of the healthy participant
and the stroke survivor are shown in Fig 9(a, c) and (b, d),
respectively. It is indicated that the 8-bar linkage mechanism
can adapt to the finger size and joint stiffness of different
participants from the following observation in the experimental
results: 1) the finger motion workspace of the stroke survivor is
smaller than that of the healthy participant because the stroke
survivor’s fingers are stiffer than the healthy participant’s
fingers; 2) the motion trajectory lengths of the stroke survivor
are slightly smaller than those of the healthy participant.
The normal active ROM of index finger for healthy subjects
are about 120◦for MCP, 100◦for PIP and 80◦for DIP,
respectively [32]–[34]. The ROM of finger joints assisted by
the hand exoskeleton are shown in Fig. 10. The results show
that the hand exoskeleton can enable 90% active range of
F/E motions for the healthy participant in average (106.5◦for
MCP, 102.5◦for PIP and 61.9◦for DIP), and 52% for the
stroke survivor in average (72.7◦for MCP, 55.1◦for PIP and
27.2◦for DIP). It is worth noting that the stroke survivor in this
study loses his capability of active F/E motion and his active
ROMs for three finger joints are all 0. To further evaluate the
effectiveness of the hand exoskeleton, we calculate the degree
8
of motion similarity between the healthy participant and the
stroke survivor by the following equation:
SI = ROMstroke
ROMhealth ×100% (14)
The degrees of motion similarity are 68.3% for MCP joint,
53.8% for PIP joint, and 43.9% for DIP joint, respectively.
The average degree of motion similarity is about 57.2%. With
more than 50% similarity, it is expected that stroke survivors
are capable of performing finger flexion and extension to some
extent and grasping objects with the assistance of the hand
exoskeleton, which is proven in Section IV-C.
(c)
(a) (b)
(d)
Fig. 9. The experimental results of finger trajectories and joint rotational
angles. The finger trajectories of the healthy participant and the stroke survivor
are shown in (a) and (b), respectively. The relationship between the rotational
angles of finger joints and of active joint (q1) of linkage mechanism with
respect to the healthy participant and the stroke survivor are shown in (c) and
(d), respectively.
Fig. 10. The comparison between the ranges of motion (ROM) of the healthy
participant and those of the stroke survivor.
C. Grasping Performance of the Hand Exoskeleton
To evaluate the grasping performance of the hand exoskele-
ton as a rehabilitation and/or assistance device for stroke
survivors, the stroke survivor was asked to accomplish the
following tasks with/without the assistance of the hand ex-
oskeleton respectively:
a) holding some large objects (such as a mineral water
bottle, a spray bottle and a tennis ball) and throwing them
into a box;
b) pinching some small objects (a table tennis ball, a
wooden block, and a die) and throwing them into a box;
c) laterally pinching some thin objects (a card and a key)
and throwing them into a box;
d) tripodally grasping a screw driver and throwing it into
a box, as well as tripodally grasping a pen and writing some
English letters.
As shown in Fig. 11 and the Supplementary Video, the
stroke survivor is able to complete all tasks with the assis-
tance of the hand exoskeleton, whereas none of them can be
accomplished without the hand exoskeleton. The experimental
results demonstrate that the proposed hand exoskeleton can
provide effective grasping assistance for stroke survivors to
perform ADLs.
D. Discussion
The hand exoskeleton in this article is made of aluminum
and textile, and its weight is only 235g. Although some rigid
hand exoskeletons [9], [25], [26], [41] weigh less than 235g,
they can only actuate a few fingers, typically the thumb and the
index finger. Most rigid five-finger exoskeletons reported in lit-
eratures have large mass, such as the hand exoskeleton in [24]
(759g). A few rigid five-finger exoskeletons weigh between
200-400g [42]–[44], which are similar to the hand exoskeleton
presented in this study. Compared with soft hand exoskeletons
[12], [13], [16], [17], the proposed hand exoskeleton has a
slightly higher weight, but it has many other advantages: (1)
it can be easily donned/doffed; (2) it can provide sufficient
grip force for patients as described in Section IV-A; (3) it can
easily adapt to the finger size and joint stiffness of different
users, ensuring that the device does not cause any harm or
injury to patients’ fingers as described in Section IV-B.
Most existing hand exoskeletons [24], [26] are typically
designed to perform motions with pre-defined trajectories,
resulting in discomfort for patients during hand operations.
On the contrary, the proposed hand exoskeleton system in this
paper is an underactuated system and does not require precise
control of every finger joint to achieve accurate trajectories for
hand motions. The linkage-based hand exoskeleton actually
provides patients a grasping assistance that is not strictly
constrained by the relationship of joint rotation angles. There-
fore, the finger motions to be assisted are not just dependent
on the motor actuation but also closely related to the finger
joint stiffness of patient users and the shape of the objects
to be grasped. Consequently, the proposed hand exoskeleton
is compatible with the natural kinematics of finger system
and adaptive to the geometry of objects, and can generate
relatively natural motions without causing any discomfort or
injury to patients. Furthermore, only one control value needs
to be estimated by the kinematical model for each finger to
grasp objects, which reduces the control complexity of the
9
()b(a) (d) ( )e
(h)
(g)
(f) (j)
(c)
(i)
Fig. 11. Examples of grasping tasks are accomplished by the stroke survivor with the assistance of the hand exoskeleton. (a) A mineral water bottle; (b)A
spray bottle; (c) A tennis ball; (d) A wooden block; (e) A die; (f) A table tennis ball; (g) A screw driver; (h) A pen; (i) A card; (j) A key.
proposed hand exoskeleton compared with those full-actuated
hand exoskeletons [45]–[47].
The proposed hand exoskeleton is secured to the patient’s
hand through Velcro straps, and the position variations of
the exoskeleton relative to the hand may be produced. Each
time a patient wears the exoskeleton, the patient needs to
slightly adjust the wearing position of the hand exoskeleton.
However, the repositioning of the hand exoskeleton doesn’t
degrade the performance of the hand exoskeleton to assist
patients’ ADLs. During the experiment, the stroke patient gave
the proposed hand exoskeleton a positive comment because it
can significantly improve the quality of his life. It is of great
significance to use the hand exoskeleton for stroke patients in
daily life.
However, the proposed hand exoskeleton system has some
limitations. Firstly, no sensors are installed on the hand
exoskeleton except the current sensors in the motors for
detecting the actuator torques. The hand exoskeleton is con-
trolled according to the pre-defined kinematic model when
patients perform hand motions. In future work, force and angle
sensors should be installed on the hand exoskeleton and a
feedback control method should be developed to improve the
autonomous performance of the grasping assistance. Secondly,
the hand exoskeleton system should be more user-friendly for
patient users to use the hand exoskeleton in daily life. A voice
control method will be developed to make patients perform
ADLs voluntarily through the hand exoskeleton. Finally, the
hand exoskeleton will be integrated with the arm exoskeleton
[48], [49] and lower limb exoskeleton [50], [51] developed in
our previous studies to better assist stroke patients to conduct
ADLs.
V. CONCLUSION
In this paper, we present an underactuated base-to-distal
hand exoskeleton for adaptive grasping assistance. A novel
underactuated 8-bar linkage mechanism is designed to actu-
ate finger flexion and extension and thumb adduction and
abduction through cables driven by remotely located mo-
tors. The base-to-distal linkage applies force only to distal
phalanges of fingers, which not only makes the hand ex-
oskeleton adapt to different sizes and stiffness of fingers but
also allows the finger phalanges to naturally accommodate
the geometry of the objects to be grasped. Furthermore, a
kinematical model of 8-bar linkage mechanism is established
to control the skeleton finger. A five-finger hand exoskeleton
with active flexion/extension (F/E) for all fingers and active
abduction/adduction (Ab/Ad) for the thumb is assembled, and
the passive adduction/abduction motion assistance of fingers
except thumb are also provided to enhance the comfort of
the hand exoskeleton, especially during hand operation. The
assistance performance of the hand exoskeleton was then
tested with a healthy subject and a stroke survivor who loses
his capability of active F/E motion completely. Experimental
results show that the hand exoskeleton can generate suffi-
cient fingertip force for regular grasping tasks. The hand
exoskeleton enables the healthy participant and the stroke
survivor to achieve 90% and 52% normal active range of
F/E motions of healthy subjects, respectively. In addition,
the stroke survivor can accomplish various hand operation
tasks with the assistance of the hand exoskeleton, such as
grasping, pinching and writing. These results demonstrate that
the underactuated base-to-distal hand exoskeleton can be an
effective device for patients with a hemiparetic hand to conduct
rehabilitation training and daily-life assistance, and, as a result,
to significantly improve their quality of life.
APPENDIX
A. The Measurement Mehtod of Finger Trajectories and Joint
Angles
To calculate the angles of the finger joints and the active
joint of the 8-bar linkage mechanism, the coordinates of finger
joints (MCP, PIP, DIP and fingertip) are defined as J1,J2,J3
and EJ, and the coordinates of mechanical joints that are
shown in Fig. 8 are defined as P1,P2,P3and P4. Then, the
vectors of three index finger phalanxes (S1,S2and S3) are
defined as follows: −→
S1=J2−J1
10
−→
S2=J3−J2
−→
S3=EJ −J3
And the vectors of the 4-bar linkage chain (L1,L2,L3and
L4) are defined as follows:
−→
L1=P2−P1
−→
L2=P3−P2
−→
L3=P4−P3
−→
L4=P1−P4
Then, all the finger joint angles can be calculated by the
following formulas:
MCP = acos ⎛
⎝−−→
S1·−→
L1
−→
S1
·
−→
L1
⎞
⎠
PIP = acos ⎛
⎝
−→
S2·−→
S1
−→
S2
·
−→
S1
⎞
⎠
DIP = acos ⎛
⎝
−→
S3·−→
S2
−→
S3
·
−→
S2
⎞
⎠
And the active joint angle q1of the 8-bar linkage mechanism
is calculated by the following formula:
q1=acos⎛
⎝−−→
L2·−→
L1
−→
L2
·
−→
L1
⎞
⎠
REFERENCES
[1] O. Alkhatib and A. Alahmar, “A literature review on length of stay
prediction for stroke patients using machine learning and statistical
approaches,” arXiv preprint arXiv:2201.00005, 2021.
[2] V. L. Feigin, M. Brainin, B. Norrving, S. Martins, R. L. Sacco, W. Hacke,
M. Fisher, J. Pandian, and P. Lindsay, “World stroke organization (wso):
Global stroke fact sheet 2022,” International Journal of Stroke, vol. 17,
no. 1, pp. 18–29, 2022.
[3] Y. Du, Y. Shi, H. Ma, D. Li, T. Su, O. Z. Meidege, B. Wang, and
X. Lu, “Application of multi-dimensional intelligent visual quantitative
assessment system to evaluate hand function rehabilitation in stroke
patients,” Brain Sciences, vol. 12, no. 12, p. 1698, 2022.
[4] A. Henderson, C. Pehoski et al.,Hand function in the child: Foundations
for remediation. Elsevier Health Sciences, 2005.
[5] F. Rehab, “Maximizing hand recovery after stroke: How to im-
prove hand and finger mobility,” 2022, https://www.flintrehab.com/
hand-recovery-after- stroke-methods/ Accessed December 12, 2022.
[6] J. Wu, H. Cheng, J. Zhang, S. Yang, and S. Cai, “Robot-assisted therapy
for upper extremity motor impairment after stroke: a systematic review
and meta-analysis,” Physical Therapy, vol. 101, no. 4, p. pzab010, 2021.
[7] S. Khalid, F. Alnajjar, M. Gochoo, A. Renawi, and S. Shimoda, “Robotic
assistive and rehabilitation devices leading to motor recovery in upper
limb: a systematic review,” Disability and Rehabilitation: Assistive
Technology, pp. 1–15, 2021.
[8] A. Chiri, F. Giovacchini, N. Vitiello, E. Cattin, S. Roccella, F. Vecchi,
and M. C. Carrozza, “Handexos: Towards an exoskeleton device for the
rehabilitation of the hand,” in 2009 IEEE/RSJ international conference
on intelligent robots and systems. IEEE, 2009, pp. 1106–1111.
[9] C. L. Jones, F. Wang, R. Morrison, N. Sarkar, and D. G. Kamper,
“Design and development of the cable actuated finger exoskeleton
for hand rehabilitation following stroke,” IEEE/Asme Transactions on
Mechatronics, vol. 19, no. 1, pp. 131–140, 2012.
[10] C. G. Rose and M. K. O’Malley, “Hybrid rigid-soft hand exoskeleton
to assist functional dexterity,” IEEE Robotics and Automation Letters,
vol. 4, no. 1, pp. 73–80, 2018.
[11] M. Sarac, M. Solazzi, E. Sotgiu, M. Bergamasco, and A. Frisoli,
“Design and kinematic optimization of a novel underactuated robotic
hand exoskeleton,” Meccanica, vol. 52, no. 3, pp. 749–761, 2017.
[12] P. Polygerinos, Z. Wang, K. C. Galloway, R. J. Wood, and C. J. Walsh,
“Soft robotic glove for combined assistance and at-home rehabilitation,”
Robotics and Autonomous Systems, vol. 73, pp. 135–143, 2015.
[13] Z. Q. Tang, H. L. Heung, K. Y. Tong, and Z. Li, “Model-based
online learning and adaptive control for a “human-wearable soft robot”
integrated system,” The International Journal of Robotics Research,
vol. 40, no. 1, pp. 256–276, 2021.
[14] J. Wang, Y. Fei, and W. Pang, “Design, modeling, and testing of
a soft pneumatic glove with segmented pneunets bending actuators,”
IEEE/ASME Transactions on Mechatronics, vol. 24, no. 3, pp. 990–
1001, 2019.
[15] Y. Wang, S. Kokubu, Z. Zhou, X. Guo, Y.-H. Hsueh, and W. Yu,
“Designing soft pneumatic actuators for thumb movements,” IEEE
Robotics and Automation Letters, vol. 6, no. 4, pp. 8450–8457, 2021.
[16] P. Tran, S. Jeong, F. Lyu, K. Herrin, S. Bhatia, D. Elliott, S. Kozin,
and J. P. Desai, “Flexotendon glove-iii: Voice-controlled soft robotic
hand exoskeleton with novel fabrication method and admittance grasping
control,” IEEE/ASME Transactions on Mechatronics, 2022.
[17] B. B. Kang, H. Choi, H. Lee, and K.-J. Cho, “Exo-glove poly ii: A
polymer-based soft wearable robot for the hand with a tendon-driven
actuation system,” Soft robotics, vol. 6, no. 2, pp. 214–227, 2019.
[18] D. H. Kim, Y. Lee, and H.-S. Park, “Bioinspired high-degrees of freedom
soft robotic glove for restoring versatile and comfortable manipulation,”
Soft Robotics, vol. 9, no. 4, pp. 734–744, 2022.
[19] S. Park, M. Fraser, L. M. Weber, C. Meeker, L. Bishop, D. Geller,
J. Stein, and M. Ciocarlie, “User-driven functional movement training
with a wearable hand robot after stroke,” IEEE Transactions on Neural
Systems and Rehabilitation Engineering, vol. 28, no. 10, pp. 2265–2275,
2020.
[20] L. Gerez, A. Dwivedi, and M. Liarokapis, “A hybrid, soft exoskeleton
glove equipped with a telescopic extra thumb and abduction capa-
bilities,” in 2020 IEEE International Conference on Robotics and
Automation (ICRA). IEEE, 2020, pp. 9100–9106.
[21] Y. Wang, W. Li, S. Togo, H. Yokoi, and Y. Jiang, “Survey on main drive
methods used in humanoid robotic upper limbs,” Cyborg and Bionic
Systems, vol. 2021, 2021.
[22] W. Chen, G. Li, N. Li, W. Wang, P. Yu, R. Wang, X. Xue, X. Zhao,
and L. Liu, “Soft exoskeleton with fully actuated thumb movements for
grasping assistance,” IEEE Transactions on Robotics, 2022.
[23] ——, “Restoring voluntary bimanual activities of patients with chronic
hemiparesis through a foot-controlled hand/forearm exoskeleton,” IEEE
Transactions on Neural Systems and Rehabilitation Engineering, 2023.
[24] W. Xu, Y. Guo, C. Bravo, and P. Ben-Tzvi, “Design, control, and
experimental evaluation of a novel robotic glove system for patients
with brachial plexus injuries,” IEEE Transactions on Robotics, 2022.
[25] G. Li, L. Cheng, Z. Gao, X. Xia, and J. Jiang, “Development of
an untethered adaptive thumb exoskeleton for delicate rehabilitation
assistance,” IEEE Transactions on Robotics, vol. 38, no. 6, pp. 3514–
3529, 2022.
[26] M. Cempini, M. Cortese, and N. Vitiello, “A powered finger–thumb
wearable hand exoskeleton with self-aligning joint axes,” IEEE/ASME
Transactions on mechatronics, vol. 20, no. 2, pp. 705–716, 2014.
[27] F. Aggogeri, T. Mikolajczyk, and J. O’Kane, “Robotics for rehabilita-
tion of hand movement in stroke survivors,” Advances in Mechanical
Engineering, vol. 11, no. 4, p. 1687814019841921, 2019.
[28] I. Sarakoglou, A. Brygo, D. Mazzanti, N. G. Hernandez, D. G. Cald-
well, and N. G. Tsagarakis, “Hexotrac: A highly under-actuated hand
exoskeleton for finger tracking and force feedback,” in 2016 IEEE/RSJ
International Conference on Intelligent Robots and Systems (IROS).
IEEE, 2016, pp. 1033–1040.
[29] J. Iqbal, H. Khan, N. G. Tsagarakis, and D. G. Caldwell, “A novel
exoskeleton robotic system for hand rehabilitation–conceptualization to
prototyping,” Biocybernetics and biomedical engineering, vol. 34, no. 2,
pp. 79–89, 2014.
[30] N. Secciani, A. Topini, A. Ridolfi, E. Meli, and B. Allotta, “A novel
point-in-polygon-based semg classifier for hand exoskeleton systems,”
IEEE Transactions on Neural Systems and Rehabilitation Engineering,
vol. 28, no. 12, pp. 3158–3166, 2020.
[31] E. Amirpour, R. Fesharakifard, H. Ghafarirad, S. M. Rezaei,
A. Saboukhi, M. Savabi, and M. R. Gorji, “A novel hand exoskeleton to
11
enhance fingers motion for tele-operation of a robot gripper with force
feedback,” Mechatronics, vol. 81, p. 102695, 2022.
[32] Roberto Grujiˇ
ci´
c MD, “Metacarpophalangeal (mcp)
joints,” 2022, https://www.kenhub.com/en/library/anatomy/
metacarpophalangeal-mcp- joints.
[33] “Normal range of motion reference values,” 2023, https://www.
eatonhand.com/nor/nor002.htm.
[34] G. Bain, N. Polites, B. Higgs, R. Heptinstall, and A. McGrath, “The
functional range of motion of the finger joints,” Journal of Hand Surgery
(European Volume), vol. 40, no. 4, pp. 406–411, 2015.
[35] S.-W. Pu, Y.-C. Pei, and J.-Y. Chang, “Decoupling finger joint motion
in an exoskeletal hand: a design for robot-assisted rehabilitation,” IEEE
Transactions on Industrial Electronics, vol. 67, no. 1, pp. 686–697, 2019.
[36] I. Jo and J. Bae, “Design and control of a wearable and force-controllable
hand exoskeleton system,” mechatronics, vol. 41, pp. 90–101, 2017.
[37] M. Barakat, J. Field, and J. Taylor, “The range of movement of the
thumb,” Hand, vol. 8, no. 2, pp. 179–182, 2013.
[38] M. Suarez-Escobar and E. Rendon-Velez, “An overview of
robotic/mechanical devices for post-stroke thumb rehabilitation,”
Disability and Rehabilitation: Assistive Technology, vol. 13, no. 7, pp.
683–703, 2018.
[39] M. H. Abdelhafiz, E. G. Spaich, S. Dosen, and L. N. A. Struijk,
“Bio-inspired tendon driven mechanism for simultaneous finger joints
flexion using a soft hand exoskeleton,” in 2019 IEEE 16th International
Conference on Rehabilitation Robotics (ICORR). IEEE, 2019, pp.
1073–1078.
[40] N. Bhatt and S. Varadhan, “Posture similarity index: a method to
compare hand postures in synergy space,” PeerJ, vol. 6, p. e6078, 2018.
[41] M. B. Hong, S. J. Kim, Y. S. Ihn, G.-C. Jeong, and K. Kim, “Kulex-
hand: An underactuated wearable hand for grasping power assistance,”
IEEE Transactions on Robotics, vol. 35, no. 2, pp. 420–432, 2019.
[42] R. Casas, M. Sandison, T. Chen, and P. S. Lum, “Clinical test of a
wearable, high dof, spring powered hand exoskeleton (handsome ii),”
IEEE Transactions on Neural Systems and Rehabilitation Engineering,
vol. 29, pp. 1877–1885, 2021.
[43] M. Sandison, K. Phan, R. Casas, L. Nguyen, M. Lum, M. Pergami-
Peries, and P. S. Lum, “Handmate: wearable robotic hand exoskeleton
and integrated android app for at home stroke rehabilitation,” in 2020
42nd Annual International Conference of the IEEE Engineering in
Medicine & Biology Society (EMBC). IEEE, 2020, pp. 4867–4872.
[44] M. Malvezzi, T. L. Baldi, A. Villani, F. Ciccarese, and D. Prattichizzo,
“Design, development, and preliminary evaluation of a highly wearable
exoskeleton,” in 2020 29th IEEE International Conference on Robot
and Human Interactive Communication (RO-MAN). IEEE, 2020, pp.
1055–1062.
[45] P. Agarwal and A. D. Deshpande, “Subject-specific assist-as-needed
controllers for a hand exoskeleton for rehabilitation,” IEEE Robotics
and Automation Letters, vol. 3, no. 1, pp. 508–515, 2017.
[46] N. Sun, G. Li, and L. Cheng, “Design and validation of a self-
aligning index finger exoskeleton for post-stroke rehabilitation,” IEEE
Transactions on Neural Systems and Rehabilitation Engineering, vol. 29,
pp. 1513–1523, 2021.
[47] G. Li, L. Cheng, and N. Sun, “Design, manipulability analysis and
optimization of an index finger exoskeleton for stroke rehabilitation,”
Mechanism and Machine Theory, vol. 167, p. 104526, 2022.
[48] N. Li, T. Yang, Y. Yang, P. Yu, X. Xue, X. Zhao, G. Song, I. H.
Elhajj, W. Wang, N. Xi et al., “Bioinspired musculoskeletal model-
based soft wrist exoskeleton for stroke rehabilitation,” Journal of Bionic
Engineering, vol. 17, no. 6, pp. 1163–1174, 2020.
[49] N. Li, T. Yang, P. Yu, J. Chang, L. Zhao, X. Zhao, I. H. Elhajj, N. Xi,
and L. Liu, “Bio-inspired upper limb soft exoskeleton to reduce stroke-
induced complications,” Bioinspiration & biomimetics, vol. 13, no. 6, p.
066001, 2018.
[50] X. Tan, B. Zhang, G. Liu, X. Zhao, and Y. Zhao, “A time-independent
control system for natural human gait assistance with a soft exoskeleton,”
IEEE Transactions on Robotics, 2022.
[51] ——, “Cadence-insensitive soft exoskeleton design with adaptive gait
state detection and iterative force control,” IEEE Transactions on Au-
tomation Science and Engineering, vol. 19, no. 3, pp. 2108–2121, 2021.
Wenyuan Chen received his B.Eng. degree in Me-
chanical Engineering and Automation from Yan-
shan University, Qinhuangdao, China, in 2017. He
is currently working toward his Ph.D. degree in
mechatronic engineering with the State Key Labora-
tory of Robotics, Shenyang Institute of Automation,
Chinese Academy of Sciences, Shenyang, China.
His current research interests include exoskeleton
robot, intelligent control and systems.
Guangyong Li (Member, IEEE) received the B.S.
degree in mechanical engineering from Nanjing Uni-
versity of Aeronautics and Astronautics, Nanjing,
China, in 1992, the M.S. degree in aerospace engi-
neering from Beijing Institute of Control Engineer-
ing, China Academy of Space Technology, Beijing,
China, in 1999, and the Ph.D. degree in electrical
engineering from Michigan State University, East
Lansing, MI, USA, in 2006.
He is currently an Associate Professor with the
Department of Electrical and Computer Engineering,
University of Pittsburgh, Pittsburgh, PA, USA. His current research interests
include micro/nanorobotic systems; fabrication of microelectromechanical
systems/nanoelectromechanical systems, nanodevices, and biosensors; control
theory and applications, real-time system design, implementation, and inte-
gration; and neural network control.
Mingwei Li received his B.Eng degree in Elec-
tronic Information Engineering from Shandong Jiao-
tong University,Shandong,China,in 2021. He is cur-
rently a research engineer with Suzhou Xianna Pre-
cision Instrument Co.Ltd, Jiangsu,China.
His current research interests include rehabilita-
tion robots, intelligent control and systems.
Xin Li received her B.Eng. degree in Electrical
Engineering and Automation from Changchun Uni-
versity of Science and Technology in 2021. She
is currently a research engineer with Suzhou Xi-
anna Precision Instrument Co. Ltd., Suzhou, Jiangsu,
China.
Her research interest includes exoskeleton appa-
ratus, robots and electronic products.
Wenxue Wang (Member, IEEE) received the B.Sc.
degree in Automatic Control from the Beijing In-
stitute of Technology, the MSc degree in Con-
trol Theory from the Institute of Systems Science,
Chinese Academy of Sciences, Beijing, and the
D.Sc. degree in Systems Science and Mathemat-
ics from Washington University in Saint Louis,
Missouri, in 1996, 1999, and 2006, respectively.
He is currently a professor at Shenyang Insti-
tute of Automation, Chinese Academy of Sci-
ences.
His research interests include robotics, bio-syncretic robotics, intelligent
control, compressive sensing, and computational and systems biology.
12
Xiujuan Xue received the M.S. degree in Rehabil-
itation Medicine from Jinzhou Medical University,
Jinzhou, China, in 2005.
She is currently a Chief Physician in the Rehabil-
itation Center for the Disabled, Shenyang, China.
Her research interests include robotic exoskeleton
robot, tactile rehabilitation of stroke, paraplegia and
children with cerebral palsy.
Xingang Zhao (Member, IEEE) received the B.E.
and M.E. dereees in mechanics from Jilin University,
Changchun, China, in 2000 and 2004, respectively,
and the Ph.D. degree in pattern recognition and
intelligent systems from the Chinese Academy of
Sciences, Shenyang, China, in 2008.
From 2015 to 2016, he was a Visiting Scien-
tist with the Rehabilitation Institute of Chicago,
Chicago, IL, USA. He is currently a Professor with
the State Key Laboratory of Robotics, Shenyang
Institute of Automation, Chinese Academy of Sci-
ences. His main research interests include medical robots, rehabilitation
robots, robot control, and pattern recognition.
Lianqing Liu (Senior Member,
IEEE) received the B.S. degree
in Industry Automation from Zhengzhou Uni-
versity, Zhengzhou, China, in 2002, and the Ph.D.
degree in pattern recognition and intelligent systems
from the Shenyang Institute of Automation, Chinese
Academy of Sciences, Shenyang, China, in 2009.
He is currently a Professor at Shenyang Institute
of Automation, Chinese Academy of Sciences. His
current research interests include mi-
cro/nanorobotics, biosyncretic robotics, and intelli-
gent control.
