Content uploaded by Min Li
Author content
All content in this area was uploaded by Min Li on Jul 03, 2020
Content may be subject to copyright.
Min Li, Tianci Wang, Yueyan Zhuo, Bo He, Tangfei Tao, Jun Xie, Guanghua Xu. A soft robotic glove for hand rehabilitation
training controlled by movements of the healthy hand. 2020. 17th International Conference on Ubiquitous Robots. Online,
Japan, June 22-26. http://www.ubiquitousrobots.org/2020/
Abstract— Most stroke patients with hand dysfunction have
normal function of one side of the body and their intact
musculoskeletal systems are intact. Their hand function can be
recovered through rehabilitation training. In this paper, a
3D-printed pneumatic-driven soft robotic glove is designed for
hand rehabilitation training controlled by the movements of the
healthy hand. Data glove is used to collect the motion data of the
healthy hand that is then used to control the robotic glove.
Characterization tests of the glove were carried out to prove the
feasibility of the soft robotic glove. The experimental results
show that the robotic glove can assist users to complete the
rehabilitation training task.
I. INTRODUCTION
Traditional hand rehabilitation training undertaken by
occupational therapists and physiotherapists are not sufficient
to cover the needs of stroke patients with hand dysfunction.
Moreover, the cost is relatively high when therapists are
involved. Methods are needed to help patients with hand
dysfunction to conduct rehabilitation training exercises by
themselves without requiring therapists’ attending. As a result,
wearable devices for hand rehabilitation are getting more and
more research attention.
In the last decade, many robotics and bioengineering
related laboratories started projects with the main purpose of
developing automatic or semi-automatic systems for hand
rehabilitation. For example, the rehabilitation robot developed
by Child et.al [1] was convenient to control and could adapt to
different sizes of fingers. Each finger of this robot was driven
independently by an actuator pulled by a wire. Nevertheless,
the control was not in real time. Moreover, the thumb design
needed further modification and wearing comfort needed to be
improved. Leonardis et al. [2] developed a rehabilitation robot
with a simple structure and the ability to detect and control the
applied force. sEMG signal of the healthy hand was used to
control the robot wearing on the affected hand. However, four
fingers of this robot could only move at the same time; the
whole structure was rigid, which has possibility to cause
secondary damage to the patient; the thumb design needed
further modification. Ueki et al. [3] developed a hand
exoskeleton hand functional rehabilitation robot driven by 15
motors. Independent extension/flexion of all five fingers and
*The research leading to these results has received funding from the
National Science Foundation of China (51975451), the Natural Science
Foundation of Shaanxi Province of China (2019JQ-332), and China
Postdoctoral Science Foundation (2019M653586).
M Li (corresponding author), T Wang, Y Zhuo, B He, T Tao, G Xu, and J
Xie are all with State Key Laboratory for Manufacturing System Engineering,
School of Mechanical Engineering, Xi’an Jiaotong University, Xi’an 710049
China (phone:+86-29-82663707;fax:86-29-82664257; e-mail:
min.li@mail.xjtu.edu.cn).
thumb's internal rotation were achieved. Virtual reality (VR)
training interface was provided. The exoskeleton robot
wearing on the affected hand can be controlled by the data
glove wearing on the healthy hand. Using this robot, patients
could carry out rehabilitation training independently. However,
the overall structure is complex. Rahman et al. [4] developed
an exoskeleton robot driven by an electric motor. The position
of each finger is detected by the resistance in the glove of the
healthy hand, so as to control the movement of the exoskeleton
robot of the affected hand. However, the exoskeleton structure
is rigid, which may cause secondary damage to the hand.
In summary, currently, the research on robots for hand
rehabilitation is mainly focused on robot structure. The
research trends include the development of exoskeletons and
wearable gloves. Most exoskeleton robots are based on rigid
links, which affect the wearing comfort and may cause further
damage to the hand. Most designs do not support the
independent movement of each finger. Robot thumb
movements are different from the hand thumb movements. To
empower patients with hand dysfunction to conduct
rehabilitation training exercises by themselves, it is a good way
to use the movements of the healthy hand to control robot
assisting the movements of the affected hand. In this paper, a
3D-printed pneumatic-driven soft robotic glove for hand
rehabilitation training controlled by the movements of the
healthy hand is presented. A data glove is used to collect finger
movement data of the healthy side of user controlling the
motion of the soft robotic glove to help the affected hand to
conduct rehabilitation training exercise. Characterization tests
of the glove are carried out to prove the feasibility of the soft
robotic glove.
II. DESIGN OF SOFT ROBOTIC GLOVE FOR HAND
REHABILITATION
A. Structural design of the soft robotic glove for hand
rehabilitation
This work is based on our previous work of a soft robotic
glove [5]. The actuator which can achieve flexing motion and
abducting motion separately. To support hand rehabilitation,
the soft actuators need to cover the range of motion of fingers
and generate adequate forces to perform daily activities. Each
human thumb has an interphalangeal (IP) joint and a
metacarpophalangeal (MCP) joint, while each of the other
fingers has a distal interphalangeal (DIP) joint, a proximal
interphalangeal (PIP) joint, and a metacarpophalangeal (MCP)
joint. Most of the time, motions of the PIP and DIP joints of
human hand are coupled, so each finger of the robotic glove
contains two separately-controlled air channels where the PIP
and DIP joints share one air channel, while the MCP joint uses
A soft robotic glove for hand rehabilitation training controlled by
movements of the healthy hand
Min Li, Tianci Wang, Yueyan Zhuo, Bo He, Tangfei Tao, Jun Xie, Guanghua Xu
Min Li, Tianci Wang, Yueyan Zhuo, Bo He, Tangfei Tao, Jun Xie, Guanghua Xu. A soft robotic glove for hand rehabilitation
training controlled by movements of the healthy hand. 2020. 17th International Conference on Ubiquitous Robots. Online,
Japan, June 22-26. http://www.ubiquitousrobots.org/2020/
one separate air channel. The MCP and IP joints of thumb are
controlled separately. The abduction motion between each
two adjacent fingers is driven by a pneumatic actuator.
The range of motion of the joint (ROM) is the movable
range when the joint moves passively or actively. Table Ⅰ
shows the normal range of finger joints from the standard of
the Japanese Society of orthopedics and the Japanese
rehabilitation Society [6]. The abduction angle between two
contiguous fingers is approximately 0-30 °. The design
requirements of a soft robotic glove is shown in Table Ⅱ
[7],[8].
TABLE I. NORMAL RANGE OF FLEXION MOTION OF FINGER JOINTS
Location
Joint
Normal
ROM(°)
Thumb
MCP
0–60
IP
0–80
The other
four
fingers
MCP
0–90
DIP
0–100
PIP
0–80
TABLE II. DESIGN REQUIREMENTS OF A SOFT ROB OTIC GLOVE
Item
Requirements
Weight of the glove
<0.5 kg
Glove size
Customizable
Total force
10–15 N
The soft robotic glove contains flexion actuators and
abduction actuators. Flexion actuators consist of several rigid
parts on the knuckles and soft parts on the joints. The rigid
parts increase finger stiffness to prevent excessive bending.
The actuator has two holes for air inflation. When the cavity is
inflated, the actuator makes deformations and helps finger
perform flexion motion because of different stiffness. Once it
is not be inflated, it return to its original state. The abduction
actuators are the same as what were used in our previous study
[5].
(a)
(b)
(d)
(c)
Figure 1. Flexion actuators and abduction actuators: (a) index flexion actuator,
(b) cross section of index flexion actuator, (c) thumb/index abduction actuator,
and (d) the other abduction actuator.
B. Fabrication
The actuators were 3D printed in a LulzBot TAZ 6
Aerostruder printer using a soft material–NinjaFlex 85A TPU.
The slice software was Cura-LulzBot 3.2.21. The printing
parameters were adjusted as Table Ⅲ.
When assembling the actuator, air tubes were glued to the
actuators using silicone sealant (RTV108, Momentive
Performance Materials, USA). Each flexion and abduction
actuators are sewn on a cloth glove.
TABLE III. PARAMETER SETTING FOR LULZBOT TAZ 6 PRINTER
Parameter
Value
lay height
0.4 mm
initial lay height
0.4 mm
line width
0.5 mm
wall thickness
0.5 mm
infill density
70%
Build Plate Temperature
55℃
printing temprature
230℃
initial printing temprature
230℃
final printing temprature
230℃
C. Rehabilitation training system
As shown in Figure 2, a data glove (WiseGlove14, Beijing
Xintian Shijing Technology, China) is used to acquire the
motion data of human hands in real time. An interface, which
is used to display the gesture and joint movement data of hand
in real time, is developed using C++ tool. A laptop computer
(ASUAfx550jx, ASUA, China) receives the joint movement
data from the data glove and sends out action commands to
analog input/output modules (JY-DAM-AIAO, Beijing Elit
Gathering Electron, China) and pressure regulators
(IVT0030-2BL, SMC, Japan) to control the air pressure in the
chambers of the soft glove. An air compressor (U-STAR601,
U-STAR, China) is used as the air source.
Data glove
Lap top
Computer
Human machine
interface
Air compressor
Pressure
regulator
Analog
input/output
module
Soft robotic
glove
Figure 2. The proposed rehabilitation training system
D. Control strategy
We assume that the MCP of ith finger (thumb to pinky is 1~5
respectively) has a flexion angle of when a pressure of
a is applied to the elastic chamber. The corresponding
control voltage is .The PIP (IP in thumb) has a flexion
angle of when a pressure of a is applied to the
elastic chamber. The corresponding control voltage is
Min Li, Tianci Wang, Yueyan Zhuo, Bo He, Tangfei Tao, Jun Xie, Guanghua Xu. A soft robotic glove for hand rehabilitation
training controlled by movements of the healthy hand. 2020. 17th International Conference on Ubiquitous Robots. Online,
Japan, June 22-26. http://www.ubiquitousrobots.org/2020/
.The abduction angle between ith and (i+1)th finger
(i=1~4) is when a pressure of a is applied to the
elastic chamber. The corresponding control voltage is .
The results of the ROM test are shown in Figures 3, 4, and 5.
The relationships between and , and ,
and are then summarized. A model to describe the
relationship between control voltage and the angle is then
established.
(3)
(6)
Note that , (j=1~4) is the fitting
coefficient.
Best on the model, the laptop computer can transform the
flexion or abduction angle from the data glove worn on
patients’ healthy hand into the voltage signal received by
pressure regulators in real time, so as to keep the flexion or
abduction angle of the actuator consistent with that of the
healthy hand.
Figure 3 The relationships between and
Figure 4 The relationships between and
Figure 5 The relationships between and
III. EXPERIMENTS AND RESULTS
Characterization tests of the glove are carried out to prove
the feasibility of the soft robotic glove. The experiment is
divided into four parts: ROM and output force test of soft
gloves, motion delay test of rehabilitation training system,
rehabilitation motion display.
A. ROM test
Figure 6 shows the test set-up to measure the ROM of the
finger actuators in the range of 0-0.3 MPA. A camera was used
to record the deformation process. The joint angles were then
measured from the recorded images. Figure 4 shows the
flexion angles. Figure 7 shows the ratio of the maximum
flexion angle to the upper limit of the normal range. By
consulting the physical therapist of Xijing Hospital, we
learned that it is effective for patients to reach the maximum
flexion angle of their fingers to 2/3 of the normal range in the
process of hand rehabilitation. As shown in Figure 8, the
actuator can reach more than 83% of upper limit of the normal
range, which prove that the flexion actuator can meet the
ROM requirements of hand rehabilitation.
Flexion actuator
Base
Air tube
Figure 6. ROM test setup
B. Output flexion force test
Figure 9 shows the set-up of the output force test. The
chamber pressure of actuator was pressurized to 0.3 MPa to
grasp the cylinder. Then the cylinder was raised by the lead
screw. The lead screw rose by 3mm each time, and the force
of the actuator was recorded by a force sensor
(BBTGTJL-1,BBTG,USA) until the actuator released from
the handle. The maximum flexion force is shown in Table IV.
The maximum force produced by flexion actuator is 36.9N,
which is larger than the required force of 15N [8]. The total
weight of the flexible gloves is 0.144kg, a 16.3% reduction
from the previous design [5].
Min Li, Tianci Wang, Yueyan Zhuo, Bo He, Tangfei Tao, Jun Xie, Guanghua Xu. A soft robotic glove for hand rehabilitation
training controlled by movements of the healthy hand. 2020. 17th International Conference on Ubiquitous Robots. Online,
Japan, June 22-26. http://www.ubiquitousrobots.org/2020/
Figure 7 . ROM of the actuator
Figure 8. Ratio of maximum flexion angle to upper limit of the normal
range
Flexion actuator
Tension sensor
Lead screw
Figure 9. Output force test platform
TABLE IV. THE MAXIMUM OUTPUT FORCE OF THE FLEXION ACTUATORS
Finger
Force
Thumb
7.700
Index
7.600
Middle
5.900
Ring
6.600
Pinky
9.100
Total
36.900
C. CPM rehabilitation training test
Continuous passive movement (CPM) rehabilitation
training, in which each finger is required to move individually
or together, is commonly used in clinic practice [9]. An
experiment was designed to feasibility of using this soft
robotic glove to conduct CPM training (see Figure 10).
A 23-year-old healthy male adult was selected as the
experimental subject. Left hand of the subject wore the soft
robotic glove, relaxing throughout the experiment; the right
hand wore a data glove to guide the left hand to complete
rehabilitation training. We selected six types of rehabilitation
actions: abduction between thumb and index finger, abduction
between index and middle finger, flexion of MCP, DIP and
PIP on each finger, thumb and index finger cross grasp, and
strong grasp (see Figure 11). The corresponding flexion angle
of each joint was recorded (Table Ⅴ).
As shown in Figure 11, the subject completed each
rehabilitation action on the training system. From Table Ⅴ, it
can be seen that the angle of each joint is within the normal
range of flexion motion of finger joints shown in Table I.
Control box
Soft robotic gloves
Human machine
interface
Data glove
Air compressor
Figure 10. The test platform of CPM rehabilitation training
Min Li, Tianci Wang, Yueyan Zhuo, Bo He, Tangfei Tao, Jun Xie, Guanghua Xu. A soft robotic glove for hand rehabilitation
training controlled by movements of the healthy hand. 2020. 17th International Conference on Ubiquitous Robots. Online,
Japan, June 22-26. http://www.ubiquitousrobots.org/2020/
(a)
(b)
(c)
(d)
(a)
(e)
(f)
(a)
Figure 11. The selected six types of rehabilitation actions in the CPM
rehabilitation training test: (a) abduction between thumb and index, (b)
abduction between index and middle, (c) flexion of MCP joint, (d) flexion of
DIP and PIP, (e) corss grasp (f) strong grasp.
TABLE V. THE FLEXION AND ABDUCTION ANGLE
Location
Joint
The actual
angle(°)
Normal
ROM(°)
Thumb
MCP
46.9
0–60
IP
50.3
0–80
abduction
45.1
0–50
The other
four
fingers
MCP
46.9
0–90
DIP
50.3
0–100
PIP
45.4
0–80
abduction
29.5
0–30
D. Grasping training test
Rehabilitation training devices are not only required to
offer CPM rehabilitation training, but also required to assist
patients to complete daily functional grasping tasks [10].With
the help of a hand rehabilitation robot, the patient can grasp
different shapes and sizes of objects, which is further helpful
to recover patients’ hand function [11]. As shown in Figure 12,
three commonly-used grasping actions were tested here: (1)
Grasping a tennis ball: a medium-sized ball, pinched with five
fingers when grasping; (2) Grasping a mark pen: small
cylinder, pinched with the thumb, index and middle fingers; (3)
Grasping a bottle: a big cylinder, holding it tightly with five
fingers when grasping. In the experiment, the subject
successfully grasped those objects using the proposed soft
robotic glove.
E. System delay test
The inherent time-lag characteristic of the
rehabilitation training system affects the user's subjective
feelings. In order to measure the system delay, the subject was
asked to repeat the strong grasp 7 times in succession, and rest
for 5 seconds in interval to relax their muscles. The whole
experimental process was recorded by a mobile phone. The
time delay between each action was read frame by frame in
the PR software. The frame rate was selected as 30 frames per
second. As is shown in Table Ⅵ, the average delay of the
system was 0.8s. Since the hand movements during
rehabilitation training are required to be slow, the time delay
would satisfy the requirement of rehabilitation
training.
(a)
(b)
(c)
Fig
ure 12. Three grasping actions used in the grasping training test: (a)grasping a
tennis ball, (b) grasping a mark pen, and (c) grasping a bottle.
TABLE VI. SYSTEM DELAY
Number
Of trials
Intervals
/frame
delay/s
average/s
1
31
1.033
0.800
2
26
0.867
3
25
0.833
4
17
0.567
5
26
0.867
6
22
0.733
7
21
0.700
IV. CONCLUSION
In this study, we designed a soft robotic glove for hand
rehabilitation training. Using the soft robotic glove, the user
can guide the hemiplegic hand to complete rehabilitation
training through a data glove worn on the healthy hand. The
characterization tests of the soft robotic glove proved that it
met the design requirements. This soft robotic glove has a
potential to assist patient to complete the rehabilitation
training tasks. In the future, further research will be carried
out in reducing system delays, improving actuator
performance, and adding haptic feedback devices and
protection devices.
REFERENCES
[1] A. Chiri, N. Vitiello, F. Giovacchini, S. Roccella, F: Veechi, and M. C.
Carrozza, "Mechatronic Design and Characterization of the Index
Finger Module of a Exoskeleton for Post-Stroke Rehabilitation," IEEE
ASME Transaction soh Mechatronics, vol. 17, no. 5, pp. 884-894, 2012.
[2] D. Leonardis et al., "An EMG-Controlled Robotic Hand Exoskeleton
for Bilateral Rehabilitation,", IEEE Transactions on Haptics, vol.8, no.
2, pp. 140-151, 2015.
Min Li, Tianci Wang, Yueyan Zhuo, Bo He, Tangfei Tao, Jun Xie, Guanghua Xu. A soft robotic glove for hand rehabilitation
training controlled by movements of the healthy hand. 2020. 17th International Conference on Ubiquitous Robots. Online,
Japan, June 22-26. http://www.ubiquitousrobots.org/2020/
[3] S. Ueki et al., "Development of a Hand-Assist Robot With
Multi-Degrees-of-Freedom for Rehabilitation Therapy," IEEE/ASME
Transactions on Mechatronics, vol. 17, no. 1, pp. 136-146, 2012
[4] M. A. Rahman and A. Al-Jumaily, "Design and Development of a Hand
Exoskeleton for Rehabilitation Following Stroke," Procedia
Engineering, vol.41, pp. 1028-1034, 2012.
[5] M. Li, Y. Zhuo, B. He, Z. Liang, G. Xu, and J. Xie, “A 3D-printed soft
hand exoskeleton with finger abduction assistance,” 16th International
Conference On Ubiquitous Robots, pp.319-322, 2019
[6] H. Di, “Mechanical design and analysis of finger rehabilitation training
device,” M.S.thesis, Dept, Mechanical, Eng., YanShan Univ, Hebei,
China, 2016
[7] 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.
[8] P. Polygerinos, K. C Galloway, E. Savage, M. Herman, K.O.Donnell
and C.J. Walsh, “Soft Robotic Glove for Hand Rehabilitation and Task
Specific Training,” IEEE International Conference on Robotics and
Automation (ICRA), pp.2913-2919,2015
[9] O'Driscoll.S.W, Giori.N.J. “Continuous passive motion (CPM): theory
and principles of clinical application”. Journal of Rehabilitation
Research and Development, vol. 37, no. 2, pp. 88-179,2000
[10] Q. Li, Li. Sun, and Z. Du, “Analysis and Research on the development
of upper limb rehabilitation robot,”(in Chinese)Journal of mechanical
design, vol. 25, no. 9, pp. 69-70, 2008
[11] S. Zhu, M. Li, G. Xu, J. Xie, B. He, C. Zhao, H. Yuan. “Finger
Exoskeleton for Rehabilitation of Finger Extension and Flexion by
Multi-Segment Mechanism,” (in Chinese) Hsi-An Chiao Tung Ta
Hsueh/Journal of Xi'an Jiaotong University, vol. 52, no. 6, pp. 17-22
and 121, June 10, 2018
