applied

sciences
Article
Design and Analysis of a Wearable Upper Limb
Rehabilitation Robot with Characteristics of
Tension Mechanism
Zaixiang Pang 1,2 , Tongyu Wang 1, *, Zhanli Wang 2 , Junzhi Yu 3 , Zhongbo Sun 4 and Shuai Liu 2
1 School of Mechatronical Engineering, Changchun University of Science and Technology,
Changchun 130022, China; pangzaixiang@ccut.edu.cn
2 School of Mechatronical Engineering, Changchun University of Technology, Changchun 130012, China;
wangzl@ccut.edu.cn (Z.W.); m15526835521@163.com (S.L.)
3 State Key Laboratory for Turbulence and Complex Systems, Department of Mechanics and Engineering
Science, BIC-ESAT, College of Engineering, Peking University, Beijing 100871, China; yujunzhi@pku.edu.cn
4 College of Electrical and Electronic Engineering, Changchun University of Technology,
Changchun 130012, China; zbsun@ccut.edu.cn
* Correspondence: wty@cust.edu.cn; Tel.: +86-1394-482-3099




Received: 12 February 2020; Accepted: 17 March 2020; Published: 20 March 2020


Abstract: Nowadays, patients with mild and moderate upper limb paralysis caused by cerebral
apoplexy are uncomfortable with autonomous rehabilitation. In this paper, according to the “rope +
toothed belt” generalized rope drive design scheme, we design a utility model for a wearable upper
limb rehabilitation robot with a tension mechanism. Owing to study of the human upper extremity
anatomy, movement mechanisms, and the ranges of motion, it can determine the range of motion
angles of the human arm joints, and design the shoulder joint, elbow joint, and wrist joint separately
under the principle of ensuring the minimum driving torque. Then, the kinematics, workspace and
dynamics analysis of each structure are performed. Finally, the control system of the rehabilitation
robot is designed. The experimental results show that the structure is convenient to wear on the
human body, and the robot’s freedom of movement matches well with the freedom of movement of
the human body. It can effectively support and traction the front and rear arms of the affected limb,
and accurately transmit the applied traction force to the upper limb of the joints. The rationality
of the wearable upper limb rehabilitation robot design is verified, which can help patients achieve
rehabilitation training and provide an effective rehabilitation equipment for patients with hemiplegia
caused by stroke.

Keywords: rehabilitation robot; characteristics of tension mechanism; wearable; upper limb; design
and analysis

1. Introduction
The number of young patients with functional impairment of the upper limbs caused by stroke
has increased rapidly, as influenced by accelerated pace of life, poor lifestyles and environmental
factors [1,2]. Limb movement disorder, which is caused by hemiplegia after stroke, not only reduces
the quality of life of patients, but also brings great pain to their physiology and psychology. Effective
rehabilitation training can improve the defect of patients’ nerve function and maintain the degree of
joint activity; it also prevents joint spasms and enhances the final rehabilitation degree of patients’
motor functions significantly [3]. The traditional rehabilitation training is one-to-one auxiliary exercise
for patients by therapists. This method is difficult to develop an effective treatment plan, and it is tough
to control accurately [4]. With the development of rehabilitation robot technology and rehabilitation

Appl. Sci. 2020, 10, 2101; doi:10.3390/app10062101 www.mdpi.com/journal/applsci

Appl. Sci. 2020, 10, 2101 2 of 22

medicine, the rehabilitation robot has become a novel motor nerve rehabilitation treatment technology.
It is of great significance to take advantage of rehabilitation robot technology for rehabilitation
training to the recovery of limb function of stroke patients [5]. The traditional methods of treatment,
which are based on the therapist’s clinical experience, have the problems of large staff consumption,
long rehabilitation cycles, limited rehabilitation effects, and so on. The research and application of
rehabilitation robot system is expected to alleviate the contradiction between supply and demand of
rehabilitation medical resources effectively, and improve the quality of life of stroke patients [6,7].
The upper limb rehabilitation robot can be divided into two types according to the structural form:
terminal traction and exoskeleton type. The terminal traction type mainly provides the rehabilitation
training of plane movement. However, the exoskeleton type extends the rehabilitation training
range from plane to three-dimensional (3D) space, which can assist the affected limb to complete
the rehabilitation training in 3D space. The exoskeleton rehabilitation robot generally drives the
movement of the patient’s limbs through the auxiliary device (also known as exoskeleton mechanical
structure). The structure of the auxiliary device is similar to the skeleton structure of the human
limbs. During the training, the patient’s limbs and the corresponding parts of the auxiliary device
are bound together, and the connecting rod of the auxiliary device swings around the corresponding
joint, so as to bring the moving limbs into motion. It can make the patient’s limbs train in different
postures through controlling the trajectories of power-assist device. At present, the structural design
method of exoskeleton rehabilitation robot is one of the hot issues in the research of rehabilitation
robot. Owing to different mechanical structures and rehabilitation principles, a variety of exoskeleton
rehabilitation robots are developed, e.g., a dynamic exoskeleton system ADEN-7 robot with 7 degrees
of freedom [8], an ARMIN robot with six degrees of freedom (four active and two passive) semi
exoskeleton structure [9], an ARMEO robot providing arm weight reduction support system training,
enhancing performance feedback and evaluation tools [10], etc. In addition, the pneumatic muscle is
used as a driver to realize four degrees of freedom active auxiliary motion RUPERT robot [11], hydraulic
drive robot LIMPACT [12], suspended rope drive robot CAREX [13]. After that, researchers developed
and designed the upper limb rehabilitation robot based on pneumatic muscle drive, unpowered upper
limb rehabilitation robot, hybrid drive upper limb rehabilitation robot and under drive exoskeleton
upper limb rehabilitation robot [14–22]. The exoskeleton rehabilitation robot solves the problem
of controlling the motion amplitude and moment of each joint of human body in the process of
rehabilitation training, and overcomes the disadvantage that the end guided rehabilitation robot can
only perform simple rehabilitation training (linear motion or circular motion) with small motion
amplitude. Currently it is a relatively safe and efficient rehabilitation robot structure. However, in
the design of exoskeleton prostheses, the matching of mechanical joint motion axis and human joint
motion axis is very important. The exoskeleton produces unexpected forces at the patient’s joint under
mismatched condition, which not only causes joint pain and injury to the patient, but also limits the
movement space of the patient’s limbs, and reduces the effect of rehabilitation training. Therefore, the
axis of each pair of motion is matched with the rotation center of each joint of the human body as far as
possible in the design of exoskeleton rehabilitation apparatus. The motion of each joint of exoskeleton
rehabilitation device is realized mainly by rotating or moving the pair, and good results have been
obtained [23,24]. Compared with the artificial rehabilitation treatment, the rehabilitation robot system
has the advantages of high training accuracy, easy to quantify the amount of exercise, and long-term
one-to-one scientific rehabilitation treatment for patients.
To satisfy the rehabilitation needs of patients with limb disorders, a wearable upper limb
rehabilitation robot is designed and developed in this article, which is mainly a device for mid-term
semi-active rehabilitation training and post-active rehabilitation training for stroke patients. Owing
to understanding the disadvantages of traditional rehabilitation training and the performances of
rehabilitation robots, combined with the human upper limb muscle anatomy characteristic and relevant
parameters, we determine the arm movement of each joint angle range from all the bones and joints of
upper limb movement characteristics, this paper proposes a design scheme of the tensegrity structure
Appl. Sci. 2020, 10, 2101 3 of 22

wearable upper limbs rehabilitation robot. The wearable upper limb rehabilitation robot is utilized
to the exercise rehabilitation treatment of hemiplegic limb to maintain the range of motion of the
limb, prevent the muscle atrophy of the limb, enhance the muscle strength of the limb, and promote
the recovery of the limb function. Therefore, it can provide an effective rehabilitation equipment for
patients with hemiplegia of upper limb caused by stroke.
In this paper, due to the study of anatomy, motion mechanism and motion range of human upper
limb, the motion angle range of each joint is determined for human arm, and the mechanical mechanism
on each degree of freedom is designed for wearable upper limb rehabilitation robot. First, to establish
the spatial pose relationship between each motion component and the end-effector of the wearable
upper limb rehabilitation robot, the motion model is established with the Denavit–Hartenberg (D-H)
parameter method and the motion space is analyzed for wearable upper limb rehabilitation robot.
The kinematics analysis is used to analyze the motion of the wearable upper limb rehabilitation robot.
Secondly, to verify whether the wearable upper limb rehabilitation robot can realize the auxiliary upper
limb functional rehabilitation training, the working space is analyzed for the wearable upper limb
rehabilitation robot. Thirdly, to analyze the output torque of wearable upper limb rehabilitation robot,
the dynamic simulation of the robot is carried out. Lastly, the control system of wearable upper limb
rehabilitation robot is designed, which obtained the tracking results of robot rehabilitation training.
It further verifies that the rationality of the design of wearable upper limb rehabilitation robot.
The main contributions of this paper are summarized as follows:
(1) Owing to the anatomy theory, motion mechanism and range of human upper limbs, a novel
wearable upper limb rehabilitation robot with tension mechanism is firstly designed, investigated
and analyzed for upper limb injured patients based on flexible transmission during rehabilitation
training process. A cable-driven modular parallel joints are innovatively designed for elbow/wrist and
a shoulder joint driven by a toothed belt. All the cable-driven motors are rear-mounted to achieve
long-distance transmission and reduce the drive inertia of the end joints. The gear belt is exploited to
drive the joints of a wearable upper limb rehabilitation robot, which realizing high precision meshing.
The design approach of the wearable upper limb rehabilitation robot facilitates the rehabilitation
training of the joint, effectively reduces the volume, mass and inertia of the actuators, and achieves the
lightweight design of the overall structure.
(2) Additionally, this paper proposes a flexibly parallel mechanism of humanoid wrist driven
by rope and supported through a compression spring. The fixed base and moving platform of the
wearable upper limb rehabilitation robot are connected by three ropes and a conical compression
spring. The springs are designed by simulating the human wrist and support the mobile platform to
complete the wrist movement, while the ropes are constructed via simulating the wrist muscles to
control the wearable upper limb rehabilitation robot. In this paper, the design approach will contribute
to the further study of parallel mechanisms with flexible joints. The results will play an important role
in reappearing the movement of human wrist and promote the development of rehabilitation robot
and rope drive technology.
(3) The kinematics and workspace of the wearable upper limb rehabilitation robot are verified
and analyzed based on the D-H method and Monte Carlo method. It demonstrates that the wearable
upper limb rehabilitation robot can satisfy the requirements of rehabilitation training through
kinematics/dynamics analysis and rehabilitation training experiments. Therefore, it also further
verifies that the feasibility and effectiveness of the design method, which provides a valuable idea for
improving rehabilitation robot mechanism.
The rest of this paper are organized as follows. From the perspective of bionics, Section 2 analyzes
the joints of the wearable upper limb rehabilitation robot, and designs the mechanical system model
of the joints. To obtain the relationship of rotation and translation between adjacent members of the
wearable upper limb rehabilitation robot and its terminal pose, the kinematics model is investigated
and analyzed for the wearable upper limb rehabilitation robot, and the correctness of the solutions
is demonstrated for forward/inverse kinematics of the wearable upper limb rehabilitation robot in
Appl. Sci. 2020, 10, 2101 4 of 22
Appl. Sci. 2020, 10, x FOR PEER REVIEW 4 of 23

correctness of the solutions is demonstrated for forward/inverse kinematics of the wearable upper
Section 3. Owing to the Monte Carlo method, the workspace and full workspace are analyzed and
limb rehabilitation robot in Section 3. Owing to the Monte Carlo method, the workspace and full
obtained for the wearable upper limb rehabilitation robot in Section 3. Section 4 simulates and analyzes
workspace are analyzed and obtained for the wearable upper limb rehabilitation robot in Section 3.
the kinematics/dynamics model of the wearable upper limb rehabilitation robot to demonstrating
Section 4 simulates and analyzes the kinematics/dynamics model of the wearable upper limb
the stability of the motion state and verifying the output torque of the motor can satisfy the needs of
rehabilitation robot to demonstrating the stability of the motion state and verifying the output
rehabilitation training process. In Section 5, the control system is designed for the wearable upper
torque of the motor can satisfy the needs of rehabilitation training process. In Section 5, the control
limb rehabilitation robot, and the rehabilitation training process is completed for upper limb injured
system is designed for the wearable upper limb rehabilitation robot, and the rehabilitation training
patients, followed by the tracking result of the robot rehabilitation training. Conclusions are drawn in
process is completed for upper limb injured patients, followed by the tracking result of the robot
Section 6.
rehabilitation training. Conclusions are drawn in Section 6.
2. Structure Design of the Wearable Upper Limb Rehabilitation Robot
2. Structure design of the wearable upper limb rehabilitation robot
2.1. Shoulder Joint Design
2.1. Shoulder Joint Design
From the anatomical point of view, the upper limb of human body mainly includes three joints:
Fromelbow
shoulder, the anatomical point ofThe
and wrist [25,26]. view, the upper
shoulder limb
joint is of human
the most body
flexible ballmainly
and socketincludes three
joint in joints:
the whole
shoulder, elbow and wrist [25,26]. The shoulder joint is the most flexible
body, which can perform flexion/extension, internal rotation/external rotation, abduction/adduction ball and socket joint in the
whole
and otherbody, which The
movements. can maximum
perform range flexion/extension, internal
of flexion/extension angle rotation/external is 135◦ ,
of shoulder jointrotation,
abduction/adduction is and ◦ , internal
135other movements. The maximum
rotation/external rotationrange
is 110of◦ , and
flexion/extension
the range of motionangle of is
shoulder
largee joint
[27]. is 135°,
At the sameabduction/adduction
time, since the three is 135°, internal
rotating pairs arerotation/external
at the beginning rotation is 110°,
of the human upperand
the range
limb motion ofchain,
motion theisoperation
largee [27].spaceAtofthethe same
terminaltime, since(hand)
effector the three
is most rotating
affected pairs aremotion
by the at the
beginning of the human upper limb motion chain, the operation space
of the joint at the beginning of the motion chain. Considering that any one degree of freedom of the terminal effector (hand)
will
is most
cause theaffected
hand toby failthe
to motion
complete ofsome
the joint at the
routine beginning
actions, of the
the three motion
degrees chain. Considering
of freedom of the shoulder that
any one
joint degree of[28].
are reserved freedom will cause
The shoulder jointthe hand to fail
is composed of to
thecomplete
glenoid ofsome routine head
the humeral actions,
andthe three
scapula,
degrees
which is of freedomby
connected ofthe
thejoint
shoulder
capsule.joint are reserved
It belongs to the[28]. The socket
ball and shoulder joint
joint. is composed
It rotates aroundof the
three
glenoid ofvertical
mutually the humeral head and scapula,
axes respectively. whichtime,
At the same is connected
the shoulderby the joint
joint capsule. Itasbelongs
is simplified to the
a connecting
ball and socket joint. It rotates around three mutually vertical axes respectively.
rod to connect the shoulder joint and the elbow joint. The degrees of freedom of shoulder joint motion At the same time,
theillustrated
is shoulder in joint is simplified
Figure 1. as a connecting rod to connect the shoulder joint and the elbow joint.
The degrees of freedom of shoulder joint motion is illustrated in Figure 1.

Figure 1.1.Shoulder
Figure Shoulderjoint freedom
joint of motion.
freedom (a) Flexion/extension;
of motion. (b) abduction/adduction;
(a) Flexion/extension; (c) internal
(b) abduction/adduction; (c)
rotation/external rotation.
internal rotation/external rotation.

The overall structure of the rehabilitation robot depends on the movement of the shoulder joint
The overall structure of the rehabilitation robot depends on the movement of the shoulder joint
and the characteristics of different body shapes. The structure of the wearable upper limb rehabilitation
and the characteristics of different body shapes. The structure of the wearable upper limb
robot shoulder joint is depicted in Figure 2. There are three degrees of freedom in the shoulder structure,
rehabilitation robot shoulder joint is depicted in Figure 2. There are three degrees of freedom in the
in which the flexion/extension and abduction/adduction of the shoulder joint are driven and connected
shoulder structure, in which the flexion/extension and abduction/adduction of the shoulder joint
by harmonic motor. The in-swing/out-swing degree of freedom with the arm as the axis cannot be used
are driven and connected by harmonic motor. The in-swing/out-swing degree of freedom with the
due to the need to wear. The arc-shaped rack on the arc-shaped slide rail is meshed with the gear of the
arm as the axis cannot be used due to the need to wear. The arc-shaped rack on the arc-shaped slide
output shaft of the reducer, and the servo motor is used as a driving force to transmit the movement of
rail is meshed with the gear of the output shaft of the reducer, and the servo motor is used as a
the input shaft of the reducer to the arc-shaped rack, so as to realize the internal/external rotation of the
driving force to transmit the movement of the input shaft of the reducer to the arc-shaped rack, so
shoulder joint. In order to ensure a certain motion accuracy, two sets of pulley blocks are installed on
as to realize the internal/external rotation of the shoulder joint. In order to ensure a certain motion
both sides of the arc-shaped rack. The function is to restrict the movement of the arc-shaped guide
accuracy, two sets of pulley blocks are installed on both sides of the arc-shaped rack. The function is
to restrict the movement of the arc-shaped guide rail along the direction of the arc, and play the role
shoulder joint internal/external rotation movement. Because the curved rack needs to cooperate
with the pulley set, the driving gear is not allowed to directly mesh with the curved rack in the
assembly space. To this end, we have added a passive gear between the active gear and the curved
rack, which can reduce the output torque of the motor reduces the quality of the mechanism and
the structure
Appl. Sci. 2020, is
10,more
2101 compact and beautiful. 5 of 22
The transmission mechanism of the shoulder joint internal/external rotation mechanism is
active gear-passive gear-arc rack, where both ends of the arc racks are provided with shoulders.
rail along the direction of the arc, and play the role of limit. Since the position of the shoulder joint
The rack gear meshes. Once it exceeds the rehabilitation range, the passive gear will be blocked by
internal/external rotation mechanism is far away from the base, the quality of the mechanism will have
the shoulder and cannot continue to move, which ensuring the safety of the patient and avoiding
a great impact on the control during the movement. After repeated experiments, it is determined that
secondary injuries to the patient.
the arc joint rack is used to achieve the shoulder joint internal/external rotation movement. Because
The structure uses gear transmission instead of belt transmission, with high transmission
the curved rack needs to cooperate with the pulley set, the driving gear is not allowed to directly mesh
accuracy, fast response speed, stable transmission and strong bearing capacity. The number of teeth
with the curved rack in the assembly space. To this end, we have added a passive gear between the
of passive gear and arc-shaped rack can limit the movement range of patient’s shoulder joint
active gear and the curved rack, which can reduce the output torque of the motor reduces the quality
rotation and ensure the safety of patients. Above all, it satisfies the requirements of high-precision
of the mechanism and the structure is more compact and beautiful.
rehabilitation training.

Figure 2. The structure diagram of shoulder joint. (a) Three-dimensional (3D) model of shoulder joint;
Figure
(b) 3D2.model
The structure diagram
of shoulder of shoulder
joint rotation degreejoint. (a) Three-dimensional (3D) model of shoulder
of freedom.
joint; (b) 3D model of shoulder joint rotation degree of freedom.
The transmission mechanism of the shoulder joint internal/external rotation mechanism is active
gear-passive
2.2. Elbow Joint gear-arc
Design rack, where both ends of the arc racks are provided with shoulders. The rack
gear meshes. Once it exceeds the rehabilitation range, the passive gear will be blocked by the shoulder
The elbow joint is the most important joint connecting the upper arm and forearm. It is a main
and cannot continue to move, which ensuring the safety of the patient and avoiding secondary injuries
trochlear joint. The joint movement is realized by the relative movement of cartilage. The flexion
to the patient.
and extension of the upper limb depend on the joint, including flexion/extension and internal
The structure uses gear transmission instead of belt transmission, with high transmission accuracy,
rotation/external rotation of the forearm [29]. These two degrees of freedom in the human body
fast response speed, stable transmission and strong bearing capacity. The number of teeth of passive
when the upper limb completes the daily activities of the movement angle: elbow joint
gear and arc-shaped rack can limit the movement range of patient’s shoulder joint rotation and ensure
flexion/extension of the maximum angle range is 135°, forearm internal rotation/external rotation is
the safety of patients. Above all, it satisfies the requirements of high-precision rehabilitation training.
90°, the largest range of motion should be considered, the elbow joint mainly completes such as
eating, holding
2.2. Elbow things, touching the head and so on in people’s daily life. If the elbow joint
Joint Design
movement is restricted, it will be greatly restricted, and other joints, which will also have a greater
The elbow joint is the most important joint connecting the upper arm and forearm. It is a
impact for the patient’s daily life. Therefore, the elbow joint plays an important role in the upper
main trochlear joint. The joint movement is realized by the relative movement of cartilage. The
limb joint. However, when the current arm is fixed internal/external rotation, there is almost no
flexion and extension of the upper limb depend on the joint, including flexion/extension and internal
internal/external rotation of shoulder joint [30]. Meanwhile, based on the redundancy of the
rotation/external rotation of the forearm [29]. These two degrees of freedom in the human body when
freedom of the motion mechanism, the above analysis of the shoulder joint has already considered
the upper limb completes the daily activities of the movement angle: elbow joint flexion/extension of
the internal/external rotation movement. To facilitate the control and the stability of the structure, it
the maximum angle range is 135◦ , forearm internal rotation/external rotation is 90◦ , the largest range
can be considered that the internal/external rotation of the forearm is fixed to move together, so that
of motion should be considered, the elbow joint mainly completes such as eating, holding things,
the internal / external rotation of the shoulder is designed to the maximum angle range of 110°. As a
touching the head and so on in people’s daily life. If the elbow joint movement is restricted, it will be
result, the flexion/extension of elbow joint is equivalent to a single joint motion pair. The degree of
greatly restricted, and other joints, which will also have a greater impact for the patient’s daily life.
freedom of elbow joint movement is shown in Figure 3a.
Therefore, the elbow joint plays an important role in the upper limb joint. However, when the current
arm is fixed internal/external rotation, there is almost no internal/external rotation of shoulder joint [30].
Meanwhile, based on the redundancy of the freedom of the motion mechanism, the above analysis of
the shoulder joint has already considered the internal/external rotation movement. To facilitate the
control and the stability of the structure, it can be considered that the internal/external rotation of the
forearm is fixed to move together, so that the internal/external rotation of the shoulder is designed to
the maximum angle range of 110◦ . As a result, the flexion/extension of elbow joint is equivalent to a
single joint motion pair. The degree of freedom of elbow joint movement is shown in Figure 3a.
Appl. Sci. 2020, 10, 2101 6 of 22
Appl. Sci. 2020, 10, x FOR PEER REVIEW 6 of 23

Figure 3.
Figure Elbow joint
3. Elbow jointstructure
structurediagram.
diagram.(a)
(a)Elbow
Elbowjoint freedom
joint of motion;
freedom (b) 3D
of motion; (b)model of elbow
3D model joint.
of elbow
joint.
As illustrated in Figure 3b, the elbow motion mechanism is constructed by two-way winding coil
structure. It avoidsinthe
As illustrated synchronization
Figure 3b, the elbow of motion
double mechanism
motor winding mechanismbybytwo-way
is constructed two-waywinding
precise
drivestructure.
coil of the motor. The driven
It avoids part of the elbowofjoint
the synchronization movement
double motor mechanism is mountedby
winding mechanism ontwo-way
the base,
and the two-way driven plate of the motor transmits the power to the two-way
precise drive of the motor. The driven part of the elbow joint movement mechanism is mounted wire plate of the elbow
on
through the rope, thus, it completes the elbow flexion/extension motion. Since the
the base, and the two-way driven plate of the motor transmits the power to the two-way wire plate elbow needs to bear
a large
of torquethrough
the elbow and has theexcellent
rope, rotational accuracy,the
thus, it completes twoelbow
crossed roller bearings motion.
flexion/extension are applied to the
Since the
elbow joint. As a consequence, it increases the bearing capacity of the elbow mechanism,
elbow needs to bear a large torque and has excellent rotational accuracy, two crossed roller bearings and reduces
the radial
are applied error of the
to the rotating
elbow joint.shaft.
As aThe use of the silicone
consequence, pad the
it increases increases
bearingthecapacity
wearingof comfort and
the elbow
aesthetics of the
mechanism, anddevice,
reduces which satisfies
the radial the of
error requirements
the rotatingofshaft.
the elbow
The usemovement when the
of the silicone padhuman arm
increases
moves and also makes it more comfortable to wear.
the wearing comfort and aesthetics of the device, which satisfies the requirements of the elbow
movement when the human arm moves and also makes it more comfortable to wear.
2.3. Wrist Joint Design
The wrist
2.3. Wrist joint is a small joint of human body, which is mainly used to connect the forearm and
Joint Design
hand, and consists of a palm and fingers. The wrist joint has two degrees of freedom to complete
The wrist joint is a small joint of human body, which is mainly used to connect the forearm
flexion/extension, abduction/adduction. When these two degrees of freedom are combined with
and hand, and consists of a palm and fingers. The wrist joint has two degrees of freedom to
pronation and supination around the long axis of forearm, the wrist joint increases the third degree of
complete flexion/extension, abduction/adduction. When these two degrees of freedom are
freedom (passive flexion/extension and extension) [31,32]. During the rehabilitation training process,
combined with pronation and supination around the long axis of forearm, the wrist joint increases
the traction of the wrist joint does not need a lot of force. In the meanwhile, the amplitude of passive
the third degree of freedom (passive flexion/extension and extension) [31,32]. During the
flexion, extension and extension of wrist in the three degrees of freedom is small. The wrist can
rehabilitation training process, the traction of the wrist joint does not need a lot of force. In the
be fully trained through the other two degrees of freedom currently. Therefore, in order to reduce
meanwhile, the amplitude of passive flexion, extension and extension of wrist in the three degrees
the complexity of the mechanical structure and restore the wrist joint of the human upper limb to
of freedom is small. The wrist can be fully trained through the other two degrees of freedom
the maximum extent, and simplify the structure and reduce the control difficulty on the premise of
currently. Therefore, in order to reduce the complexity of the mechanical structure and restore the
ensuring the basic functions, the passive flexion and extension and extension are not included in the
wrist joint of the human upper limb to the maximum extent, and simplify the structure and reduce
design requirements of the wearable upper limb rehabilitation robot described in this paper. When
the control difficulty on the premise of ensuring the basic functions, the passive flexion and
the upper limb of human body completes daily activities, the maximum range of flexion/extension
extension and extension are not included in the design requirements of the wearable upper limb
angle is 150◦ , and the abduction/adduction is 50◦ . The wrist joint can be equivalent to a spherical hinge
rehabilitation robot described in this paper. When the upper limb of human body completes daily
mechanism. Based on the inherent rigidity of the general mechanical structure, and better map of the
activities, the maximum range of flexion/extension angle is 150°, and the abduction/adduction is 50°.
movement structure of the human wrist, the wrist of the wearable upper limb rehabilitation robot is
The wrist joint can be equivalent to a spherical hinge mechanism. Based on the inherent rigidity of
designed with a flexible structure tower spring. In this paper, people mainly complete such actions as
the general mechanical structure, and better map of the movement structure of the human wrist,
eating, taking things and touching their heads in the daily life. The wrist joint not only has a high
the wrist of the wearable upper limb rehabilitation robot is designed with a flexible structure tower
frequency of motion, but also is the part of the upper limb that bears the largest load in the process
spring. In this paper, people mainly complete such actions as eating, taking things and touching
of supporting, pushing and pulling. The design of wearable upper limbs rehabilitation robot mainly
their heads in the daily life. The wrist joint not only has a high frequency of motion, but also is the
for medium-term and semi-active rehabilitation training in patients with cerebral apoplexy and late
part of the upper limb that bears the largest load in the process of supporting, pushing and pulling.
active rehabilitation training device, furthermore, the wrist of patients should have certain activity.
The design of wearable upper limbs rehabilitation robot mainly for medium-term and semi-active
During rehabilitation training, patients need to hold the end adjusting grip of the wearable upper limb
rehabilitation training in patients with cerebral apoplexy and late active rehabilitation training
rehabilitation robot, and the upper limb follows the robot to do corresponding rehabilitation training.
device, furthermore, the wrist of patients should have certain activity. During rehabilitation
Figure 4a shows the degree of freedom of wrist movement.
training, patients need to hold the end adjusting grip of the wearable upper limb rehabilitation
robot, and the upper limb follows the robot to do corresponding rehabilitation training. Figure 4a
shows the degree of freedom of wrist movement.
 Appl.Sci.
Appl. Sci.2020,
2020,10,
10,2101
x FOR PEER REVIEW 77ofof22
23

Figure 4. Wrist joint structure diagram. (a) Wrist joint freedom of motion; (b) 3D model of wrist joint.
Figure 4. Wrist joint structure diagram. (a) Wrist joint freedom of motion; (b) 3D model of wrist
As shown in Figure 4b, a flexible parallel mechanism
joint. is proposed to simulate the human wrist
with a rope drive. The wrist adopts the hand-wrist-forearm connection. The front and rear sections of
the wristAs shown in Figure
are connected by4b, a flexible
a tapered parallel mechanism
compression spring, which is proposed
is usedto tosimulate
simulatethe thehuman
motionwrist of a
with a rope drive. The wrist adopts the hand-wrist-forearm connection.
wrist joint. There are three sets of rope mechanism around, each set of rope mechanism is separated by The front and rear sections
of ◦the
120 to wrist
simulateare connected by a tapered
the wrist muscles, which compression
complete the spring,
drivewhich is used of
and control to the
simulate
wrist.the In motion
addition, of
a wristmechanism
control joint. There are three with
is equipped sets aofpower
rope source
mechanism around,
to be placed each
in the basesetpart.
of rope mechanism is
separated
Owingby to120°
loadto andsimulate
deformation the wristare muscles,
nonlinear, which complete the
and comparing withdrive and control
a cylindrical of the
helical wrist.
spring,
In addition,
the conical helical control mechanism
spring has a greateris equipped
stabilitywith and aprevents
power sourceresonance to bephenomenon,
placed in thewhich base part.
is applied
more Owing
and more to load
widely.andSpecifically,
deformation are load
if the nonlinear,
does notandmakecomparing
the spring withcoil a cylindrical
contact, thehelical spring,
relationship
the
of theconical
load and helical spring has
deformation a greater
is linear, and stability
if the load and prevents
continues to resonance
increase, then phenomenon, which is
the spring contacts
applied
from more
a large ring,and more
and the widely.
relationship Specifically,
of load and if the load does is
deformation not make the spring coil contact, the
nonlinear.
relationship
The flexible of the loadmechanism
parallel and deformation takes the is human
linear, wrist
and ifasthetheload
bionic continues
object, where to increase,
the fixedthen ringtheis
spring contacts
equivalent to thefrom
radius a large ring, complex,
and ulnar and the relationship
and the moving of loadringand deformation
is equivalent is nonlinear.
to the metacarpal bone.
The flexible
The driving rope and parallel
springmechanism
represent the takes the human
muscles wrist asaround
and ligaments the bionic object,
the wrist where thewhich
respectively, fixed
ring is equivalent to the radius and ulnar complex, and the
providing kinetic energy and support for the motion of the radial and middle wrist joints. The parallel moving ring is equivalent to the
metacarpal uses
mechanism bone.threeThe servo
driving rope to
motors anddrivespring
threerepresent
ropes, which the muscles
realize the andwrist
ligaments around the
flexion/extension
wrist
and respectively,
ulnar/radial which providing
movement of the robot. kinetic
It canenergy
not onlyandreduce
support thefor the motion
flexible degreeofofthe radial but
freedom, and
middle
also enhance wristthe joints. The of
stability parallel mechanism
the mechanism, and uses
make threetheservo
mechanismmotorssatisfyto drivethe three
motion ropes, which
amplitude
realize
of the wrist the wrist
underflexion/extension
different angles when and ulnar/radial
the mechanism movement of the robot.
is in retraction It can not flexing
and abduction, only reduceand
the flexible degree of freedom, but also enhance the stability
stretching. Therefore, the mechanism can achieve the wrist joint adduction and abduction, flexion of the mechanism, and make the
mechanism
straight action.satisfy
The the motion
flexible wristamplitude
joint driven of thebywrist
rope is under
mainly different
composed angles of when the mechanism
three parts: adjustableis
in retraction
grip, and abduction,
flexible parallel mechanism flexing
andandforearmstretching.
fixation. Therefore, the mechanism can achieve the wrist
jointAccording
adductiontoand abduction,
above analyses flexion
of thestraight
motionaction.
freedom Theofflexible
each joint wrist of joint drivenlimb,
the upper by ropea sixis
mainly composed
degrees of freedomofwearable
three parts: adjustable
upper grip, flexible parallel
limb rehabilitation robot with mechanism and forearmoffixation.
the characteristics tension
According
mechanisms to above in
is proposed analyses
this paper.of the The motion freedomofoffreedom
six degrees each joint of ofthethe upper limb,robot
rehabilitation a six
degreesthree
include of freedom
degrees wearable
of freedom upper
of thelimb rehabilitation
shoulder robot with the internal/external
joint: flexion/extension, characteristics ofrotation, tension
mechanisms is proposed
abduction/adduction; one in this paper.
degree of freedom The six degrees
of the elbowofjoint:
freedom of the rehabilitation
flexion/extension; one degreerobot
include three degrees of freedom of the shoulder joint: flexion/extension,
of freedom of the forearm: internal/external rotation; two degrees of freedom of the wrist joint: internal/external rotation,
abduction/adduction;
flexion/extension one degree of freedom
and abduction/adduction. Theof3D the elbow joint:
structure of theflexion/extension;
rehabilitation robot one degree in
is shown of
freedom
Figure of the
5. The forearm:
wearable upperinternal/external
limb rehabilitation rotation;
robot two degreesin of
designed thisfreedom
article isofwearable,
the wristwhich joint:
flexion/extension
consists of a vertical andwearable
abduction/adduction.
rehabilitationThe robot3Dandstructure
a seat:of thethe patient
rehabilitation
sits onrobotthe seatis shown
during in
Figure
the 5. The wearable
rehabilitation process; upper limb rehabilitation
the patient’s arm passesrobot through designed in thisofarticle
the traction is wearable, which
the rehabilitation robot
consists of acontacts.
institutional vertical wearable rehabilitation
The arm passively performsrobotfull-circle
and a seat: the patient
rotation sits onthe
by driving thecurved
seat during
rack and the
rehabilitation
provides resistanceprocess;
during the patient’s
active arm passes
movements through the
of the shoulders. Thetraction
seat height of the rehabilitation
adjustment mechanismrobot
institutional
can be adjusted contacts.
according Thetoarm thepassively
height of performs
the upperfull-circle
limb and rotation
the bodyby driving
shape of thethehuman
curvedbody, rack
and provides resistance during active movements of the shoulders.
adapt to different treatment environment and the difference of the patient’s body shape, and ensure The seat height adjustment
mechanism
that the affected canlimbbe adjusted
is trainedaccording to theplane
on the sagittal height of thethe
during upper limb andprocess.
rehabilitation the body shape of the
human body, adapt to different treatment environment and the difference of the patient’s body
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 23

shape,
Appl. Sci. and
2020, ensure
10, 2101 that the affected limb is trained on the sagittal plane during the rehabilitation
8 of 22
process.

Figure 5. A 3D model of the wearable upper limb rehabilitation robot.

Figure 5. A 3D model of the wearable upper limb rehabilitation robot.
Each joint mechanism of the wearable upper limb rehabilitation robot is closely combined with the
upperEacharm,joint mechanism
forearm of the
and wrist of thewearable
humanupper limbthe
body, and rehabilitation
rehabilitationrobot is closely
training combined
of multiple with
joints of
the upper
the upperlimb
arm,isforearm
realized and wristthe
by using of freedom
the human body,
of the and the rehabilitation
rehabilitation robot hinge type training of multiple
rotating pair, arc
jointstower
rack, of thespring,
upperetc.limb
Onistherealized
sagittalby using
plane, thethe freedom ofrobot
rehabilitation the rehabilitation
can realize the robot hinge type
flexion/extension
of the shoulder 0◦ –100◦ , abduction/adduction 0◦ –120◦ , internal/external rotation out 0◦can
rotating pair, arc rack, tower spring, etc. On the sagittal plane, the rehabilitation robot –110realize
◦ ; the
the flexion / extension of the shoulder
◦ ◦ 0°–100°, abduction / adduction ◦ 0°–120°,
◦
flexion/extension of the elbow 0 –105 , internal/external rotation out 0 –90 ; the flexion/extension internal / external
0rotation outmotion
◦ –90◦ joint 0°–110°;analysis
the flexion
angle / extension of the
range of the wrist elbow
meets0°–105°, internal
the angle / external
requirements ofrotation
the upperoutlimb
0°–
90°; the flexion
rehabilitation / extension
training 0°–90°
process [33]. joint motion
The free analysis angle
distribution range
of human of the
upper wrist
limbs andmeets the angle
rehabilitation
requirements of the upper limb rehabilitation
robot degrees of freedom are shown in Table 1. training process[33]. The free distribution of human
upper limbs and rehabilitation robot degrees of freedom are shown in Table 1.
Table 1. Free distribution of human upper limbs and robot degrees of freedom.
Table 1. Free distribution of human upper limbs and robot degrees of freedom.
Movement Range of Robot
Parts Degrees of Freedom Movement Range of Human Movement
Parts Degrees of freedom Movement range of human [34,35]range of
robot[34,35]
Shoulder Flexion/extension 0◦ –90◦ /0◦ –45◦ 0◦ –90◦ /0◦ –10◦
Shoulder
Shoulder Flexion/extension
Abduction/adduction 00°–90°/0°–45°
◦ –90◦ /0◦ –45◦ 00°–90°/0°–10°
◦ –75◦ /0◦ –45◦

Shoulder
Shoulder Abduction/adduction
Internal/external rotation ◦ –80◦ /0◦ –30◦
00°–90°/0°–45° ◦ –80◦ /0◦ –30◦
00°–75°/0°–45°
Elbow Flexion/extension
Internal/external 0◦ –135◦ 0◦ –105◦
Shoulder
Elbow Internal/external rotation 00°–80°/0°–30°
◦ –45◦ /0◦ –45◦ 00°–80°/0°–30°
◦ –45◦ /0◦ –45◦
rotation
Wrist Flexion/extension 0◦ –75◦ /0◦ –75◦ 0◦ –45◦ /0◦ –45◦
Elbow Flexion/extension 0°–135° 0°–105°
Internal/external
Elbow
3. Kinematic Analysisrotation 0°–45°/0°–45° 0°–45°/0°–45°

Wrist Flexion/extension
Because the object 0°–75°/0°–75°
of robot’s service is the injured limb, the injured limb wears 0°–45°/0°–45°
on the robot and
moves together under its traction to achieve rehabilitation training. It is a basis of motion control and
3. Kinematic
execution Analysis training. In order to enable the rehabilitation robot to perform more efficient
of rehabilitation
motion control
Because in object
the the process of rehabilitation
of robot’s training,
service is the injuredthe movement
limb, between
the injured limb the
wearsrobot’s endrobot
on the and
each joint can be coordinated by establishing the spatial pose relationship between the
and moves together under its traction to achieve rehabilitation training. It is a basis of motion robot’s motion
components
control and and the end-effector.
execution The movement
of rehabilitation training.variation
In orderof each joint ofthe
to enable the rehabilitation
wearable rehabilitation
robot to
robot can be appropriately changed, and the movement between the end of the wearable
perform more efficient motion control in the process of rehabilitation training, the movement rehabilitation
robot and the
between eachrobot’s
joint can be adjusted
end and eachto joint
achieve
canthebeexpected rehabilitation
coordinated training requirements.
by establishing the spatial pose
relationship between the robot’s motion components and the end-effector. The movement variation
3.1. Forward Kinematics
of each joint of the wearable rehabilitation robot can be appropriately changed, and the movement
The wearable
between the end ofupper limb rehabilitation
the wearable robot
rehabilitation is aand
robot typical
eachhuman-machine cooperation
joint can be adjusted system.
to achieve the
The robot is consistent with the movement
expected rehabilitation training requirements. of the affected limb of the human body. Therefore, to
accurately obtain the motion curve of the affected limb, a forward kinematic analysis is required
for
3.1.the wearable
Forward upper limb rehabilitation robot. To ensure that the designed wearable upper limb
Kinematics
rehabilitation robot has good applicability and practical applications, the D-H parameter model of
the wearable upper limb rehabilitation robot based on the D-H coordinate system method needs to
the wearable
accurately upper
obtain the limb
motion rehabilitation
curve of therobot. affectedTolimb,ensure that thekinematic
a forward designedanalysis
wearable upper limb
is required for
rehabilitation
the wearable robot upperhas limbgood applicability
rehabilitation and practical
robot. To ensure applications,
that the designed the D-Hwearable
parameter model
upper limbof
the wearable upper limb rehabilitation robot based on the D-H
rehabilitation robot has good applicability and practical applications, the D-H parameter model of coordinate system method needs to
set
the related
wearable parameters,
upper limb which includingrobot
rehabilitation describing
based the on theconnecting
D-H coordinate rod, which usedmethod
system to describeneedsthe to
geometric
set related characteristic
parameters, parameters
which including of describing
connecting the rods, the connection
connecting rod, parameter
which used to relationship
describe the
Appl. Sci. 2020, 10, 2101 9 of 22
between
geometrictwo connecting rods
characteristic and theofparameters
parameters connecting that define
rods, thethe relationship
connection betweenrelationship
parameter connecting
rods.
between Lastly,
two the parameters
connecting rodscanandbethe brought
parameters into the thatcorrelation
define the transformation
relationship between matrix connecting
to get the
corresponding
set results by setting
which the parameters.
rods. Lastly, the parameters can be brought into the correlation transformation matrix to get the
related parameters, including describing the connecting rod, which used to describe the
In order
geometric
corresponding to obtain
characteristic
results the relationship
parameters
by setting theofparameters.
connectingbetween rods,the therotation
connection and translation
parameter among adjacent
relationship between
members
two connecting
In orderof the wearable
torods and the
obtain theupper
parameterslimbthat
relationship rehabilitation
define the
between robot
therelationship
rotation andand its translation
betweenterminal pose,
connectingamong a reference
rods. Lastly,
adjacent
coordinate
the parameters system can should
be be
brought assigned
into the to each
correlation joint [36].
transformation
members of the wearable upper limb rehabilitation robot and its terminal pose, a reference In this paper,
matrix according
to get the to kinematics
corresponding
theory,
results
coordinate Denavit–Hartenberg
by setting
system theshould be(D-H)
parameters. assignedcoordinate
to each system
joint [36]. methodIn thisis paper,
adopted to establish
according the D-H
to kinematics
parameter
In order
theory, model of thethe
to obtain
Denavit–Hartenberg upper limb rehabilitation
relationship
(D-H) between the
coordinate robot
rotation
system ontology (see
and translation
method Figure
is adopted 6),
to as
among shown in
adjacent
establish theFigure
membersD-H
7.
of The
the geometric
wearable size
upper of
limbeach connecting
rehabilitation rod
robot of
and the rehabilitation
its terminal
parameter model of the upper limb rehabilitation robot ontology (see Figure 6), as shown in Figure pose, robot
a can
referencebe described
coordinate by four
system
should
7. The be geometric
parameters. ai and
assigned size αofi each
to each jointused
are [36]. to
In describe
connectingthis paper,
rod of theaccording
the to kinematics
rehabilitation
geometric features robotoftheory, Denavit–Hartenberg
can connecting
the be described rod byitself.
four
parameters. ai and αi are used to describe the geometric features of the connecting rod itself.
(D-H) coordinate system method is adopted to establish the D-H parameter model of the upper limb
The numericalrobot
rehabilitation values are determined
ontology (see Figure by 6), astheshown
distance and included
in Figure angle between
7. The geometric size of each axes zi −1
the connecting
αi are axes zthe
and zi The other parameters offset d i and joint angle θ i represent the connection relationship
rod
The of the rehabilitation
numerical values are robot can be described
determined by four parameters.
by the distance and included ai and
angle used to
between thedescribe i −1
geometric features of the connecting rod itself. The numerical values are determined by the distance
between
and zi The
and includedthe other
twoangle parameters
connecting
betweenrods, the anddthe
offset
axes and
zi i−1values
and joint
ziare angle
The θi parameters
determined
other represent the
by the distance connection
offset diand relationship
andincluded
joint angle
angle θi
between the
represent
between the
the two
axes xi −1 and
connecting
connection xi .The
rods,
relationship and the values
between
parameters the ofarethe
two determined
connecting
rehabilitation by the
rods, anddistance
robot the
D-H and
values
model included
are angle
determined
are listed in
2. the axes xi −1 and xi .The parameters of thei−1rehabilitation
by the
between distance and included angle between the axes x and xi .The parameters of the rehabilitation
robot D-H model are listed in
Table
robot D-H model are listed in Table 2.
Table 2.

Figure 6. 3D rehabilitation robot.

Figure 6. 3D rehabilitation robot.
Figure 6. 3D rehabilitation robot.

Figure 7. Denavit–Hartenberg (D-H) parameter model of robot.

Figure 7. Denavit–Hartenberg (D-H) parameter model of robot.
D-H parameters of upper
Table7.2.Denavit–Hartenberg
Figure (D-H)limb rehabilitation
parameter model of robot.
robot.
Table 2. D-H parameters of upper limb rehabilitation robot.
Joint i Length of Bar ai Knob Angle α Offset di Joint Angle θi Angle Range
2. D-H parameters ofiupper limb rehabilitation
Joint i LengthTable
of bar ai Knob angle αi Offset di Joint angle robot.
θi Angle range
1 0 90 ◦ 235 mm θ1 (−90 )◦ −90◦ –30◦
Joint
2 i Length0 of bar ai Knob 90angle αi Offset
◦ 20 mmdi Joint θ2angle
(−90 )θi Angle
◦ −90range
◦ –45◦

3 0 90 ◦ 420 mm θ3 (−90 )◦ −80◦ –45◦

4 0 90 ◦ 0 θ4 (0 )
◦ 0◦ –135◦
5 0 90◦ 265 mm θ5 (0◦ ) −45◦ –45◦
6 83 mm 0◦ 0 θ6 (−90◦ ) −45◦ –45◦

According to the spatial coordinate system established for each joint of the wearable upper limb
rehabilitation robot, forward kinematics analysis is carried out, i.e., the coordinate transformation from
Appl. Sci. 2020, 10, 2101 10 of 22

coordinate system {Oi−1 } to coordinate system {Oi }. Based on the robotics theory, the vector described
in coordinate system {Oi } is mapped to coordinate system {Oi−1 } by coordinate transformation:

i−1 T
i = Ai = Rot(Z, θi ) × Trans(0, 0, di ) × Trans(ai , 0, 0) × Rot(X, ai )
   
 cθi −sθi 0 0  1 0 0 ai  1 0 0 0 
   
 sθ cθi 0 0  0 1 0 0  0 cθi −sαi 0 
=  i
 0 0 1 0  0 0 1 di  0 sαi cαi 0 
  

0 0 0 1 0 0 0 1 0 0 0 1 (1)
  
 
 cθi −sθi cαi sθi sαi ai cθi 
 sθ cθi cαi −cθi sαi ai sθi 
=  i

 0 sαi cαi di 


0 0 0 1


where c means a cosine function and s represents sine function.

For a wearable upper limb rehabilitation robot, when the coordinate system of each link is
determined, the parameters of each link can be obtained. According to Formula (1), it can be known
that the pose matrix Ai between the two rods is:
 
 cθ1 0 sθ1 0 
 
 sθ1 0 −cθ1 0
A1 = Rot(Z, θ1 ) × Trans(0, 0, d1 ) × Trans(a1 , 0, 0) × Rot(X, α1 ) = 


 0 1 0 235 


0 0 0 1


 
 cθ2 0 sθ2 0 
 
 sθ2 0 −cθ2 0 
A2 = Rot(Z, θ2 ) × Trans(0, 0, d2 ) × Trans(a2 , 0, 0) × Rot(X, α2 ) =  
 0 1 0 20 

0 0 0 1


 
 cθ3 0 sθ3 0 
 
 sθ3 0 −cθ3 0 
A3 = Rot(Z, θ3 ) × Trans(0, 0, d3 ) × Trans(a3 , 0, 0) × Rot(X, α3 ) =  
 0 1 0 420 


0 0 0 1


 
 cθ4 0 sθ4 0 
 sθ 0 −cθ 0 
4 4
A4 = Rot(Z, θ4 ) × Trans(0, 0, d4 ) × Trans(a4 , 0, 0) × Rot(X, α4 ) = 
 
 0 1 0 0


 
0 0 0 1
 
 cθ5 0 sθ5 0 
 
 sθ5 0 −cθ5 0 
A5 = Rot(Z, θ5 ) × Trans(0, 0, d5 ) × Trans(a5 , 0, 0) × Rot(X, α5 ) =  
 0 1 0 265 


0 0 0 1


 
 cθ6 −sθ6 0 83cθ6 
 sθ cθ6 0 83sθ6 

6
A6 = Rot(Z, θ6 ) × Trans(0, 0, d6 ) × Trans(a6 , 0, 0) × Rot(X, α6 ) = 

 0 0 1 0 


0 0 0 1


Therefore, the matrix transformation calculation formula can be obtained by the robot’s end
handle posture relative to the robot’s base coordinate system:
Appl. Sci. 2020, 10, 2101 11 of 22

 
 r11 r12 r13 Px 
 
0
 r21 r22 r23 Py 
6 T = A 1 A 2 A 3 A 4 A 5 A 6 = 
 r
 (2)
 31 r32 r33 Pz 

0 0 0 1
 

 
 r11 r12 r13  h iT
 
where  r21 r22 r23  is the direction vector of the terminal, and, px p y pz is the position
 
r31 r32 r33
vector of the terminal.
To obtain the solution of positive kinematics, the matrices are multiplied as follows:

r11 = sθ6 (s θ4 (s θ1 sθ3 +cθ1 cθ2 cθ3 ) − cθ1 cθ4 sθ2 ) − cθ6 (s θ5 (c θ3 sθ1 −cθ1 cθ2 sθ3 )
−cθ5 (c θ4 (s θ1 sθ3 +cθ1 cθ2 cθ3 ) + cθ1 sθ2 sθ4 ))

r12 = sθ6 (s θ5 (c θ3 sθ3 −cθ1 cθ2 sθ3 ) − cθ5 (c θ4 (s θ1 sθ3 +cθ1 cθ2 cθ3 ) + cθ1 siθ2 sθ4 ))
+cθ6 (s θ4 (s θ1 sθ3 +cθ1 cθ2 cθ3 ) − cθ1 cθ4 sθ2 )
r13 = cθ5 (c θ3 sθ3 −cθ1 cθ2 siθ3 ) + sθ5 (c θ4 (s θ1 sθ3 +cθ1 cθ2 cθ3 ) + cθ1 sθ2 sθ4 )
r21 = cθ6 (s θ6 (c θ1 cθ3 +cθ2 sθ1 sθ3 ) − cθ5 (c θ4 (c θ1 sθ3 −cθ2 cθ3 sθ1 ) − sθ1 sθ2 sθ4 ))
−sθ6 (si θ4 (c θ1 sθ3 ) − cθ2 cθ3 sθ1 ) + cθ4 sθ1 sθ2 )

r22 = − sθ6 (s θ5 (c θ1 cθ3 +cθ2 sθ1 sθ3 ) − cθ5 (c θ4 (c θ1 sθ3 −cθ2 cθ3 sθ1 ) − sθ1 sθ2 sθ4 ))
−cθ6 (s θ4 (c θ1 sθ3 −cθ2 cθ3 sθ1 ) + cθ4 sθ1 sθ2 )
r23 = −cθ5 (c θ1 cθ3 +cθ2 sθ1 sθ3 ) − sθ5 (c θ4 (c θ1 sθ3 −cθ2 cθ3 sθ1 ) − sθ1 sθ2 sθ4 )

r31 = sθ6 (c θ2 cθ4 +cθ3 sθ2 sθ4 ) − cθ6 (c θ5 (c θ2 sθ4 −cθ3 cθ4 sθ2 ) − sθ2 sθ3 sθ5 )

r32 = cθ6 (c θ2 cθ4 +cθ3 sθ2 sθ4 ) + sθ6 (c θ5 (c θ2 sθ4 −cθ3 cθ4 sθ2 ) − sθ2 sθ3 sθ5 )

r33 = −sθ5 (c θ2 sθ4 −cθ3 cθ4 sθ2 ) − cθ5 sθ2 sθ3

Px = 20sθ1 + 265sθ4 (sθ1 sθ3 + cθ1 cθ2 cθ3 )
−83cθ6 (sθ5 (cθ3 sθ1 − cθ1 cθ2 sθ3 ) − cθ5 (cθ4 (sθ1 sθ3 + cθ1 cθ2 cθ3 ) + cθ1 sθ2 sθ4 ))
+83sθ6 (sθ4 (sθ1 sθ3 + cθ1 cθ2 cθ3 ) − cθ1 cθ4 sθ2 ) + 420cθ1 sθ2 − 265cθ1 cθ4 sθ2

P y = 420sθ1 sθ2 −20cθ1 −265sθ4 (c θ1 sθ3 −cθ2 cθ3 sθ1 ) + 83cθ6 (s θ5 (c θ1 cθ3 +cθ2 sθ1 sθ3
−cθ5 (c θ4 (c θ1 sθ3 −cθ2 cθ3 sθ1 ) − sθ1 sθ2 sθ4 )) − 83sθ6 (s θ4 (c θ1 sθ3 −cθ2 cθ3 sθ1 ) + cθ4 sθ1 sθ2 ) − 265cθ4 sθ1 sθ2

Pz = 83sθ6 (c θ2 cθ4 +cθ3 sθ2 sθ4 ) − 420cθ2 −83cθ6 (c θ5 (c θ2 sθ4 −cθ3 cθ4 sθ2 ) − sθ2 sθ3 sθ5 )
+265cθ2 cθ4 +265cθ3 sθ2 sθ4 +235

3.2. Workspace Analysis

The workspace of the rehabilitation robot is the set of points that the hand reference points can
reach in space during the operation of the rehabilitation robot [37]. It is a key element to estimate
the robot rehabilitation training. Moreover, it provides a key point for evaluating the rationality of
machine design. In this paper, the calculation is the set of the end can reach the target point when the
rehabilitation robot is given all pose, i.e., the full workspace. The Monte Carlo method is exploited
to analyze the workspace of the rehabilitation robot, being a kind of numerical calculation method
guided by probability and statistics theory. Taking the end of the wearable upper limb rehabilitation
robot as the reference point and combining the position vector in the forward kinematics equation, the
set of all random points that generate the reference point constitutes the workspace of the wearable
upper limb rehabilitation robot. In the specified range, the angle value of each joint is generated by
random sampling method, and then the angle value is substituted into the positive solution calculation
Appl. Sci. 2020, 10, 2101 12 of 22

program to obtain the corresponding end position point. After 100,000 simulation computations, which
can be obtained the terminal spot cloud.
The Programming and Robotics Toolbox of MATLAB software is utilized to adjust the random
points. When the sample size is 100,000, the simulation diagram of the wearable upper limb
rehabilitation
Appl. Sci. 2020, 10,robot’s workspace
x FOR PEER REVIEWis obtained, as shown in Figure 8. 13 of 23

Figure 8. A
Figure 8. A3D3Dworkspace
workspacesimulation of of
simulation thethe
robot: (a) (a)
robot: 3D 3D
spatial diagram;
spatial (b) XOY
diagram; view;view;
(b) XOY (c) XOZ
(c) view;
XOZ
(d) YOZ
view; (d)view.
YOZ view.

Geometric
3.3. Inverse construction can be calculated more easily to construct the boundary of the maximum
Kinematics
range. Due to the linear relationship between the components, the terminal pose corresponding to
Wearable
the joint angles upper
cannotlimb rehabilitation
be predicted robots
in extreme often
cases. assist patients
However, withmethod
this design hemiplegia to complete
is to simulate the
some tasks in daily life, such as eating, holding things,
possible arrival space, and give the relevant probability point cloud.and touching the head, to achieve the
purpose of restoring
As can upper
be seen from limb
Figure 8, motor function.isIfa the
the workspace task
sector in is
XOZgiven, theand
view, end position
it is is reached
approximate by
in XOY
the wearable upper limb rehabilitation robot at this time. In order to determine the parameters of
view and YOZ view, which is roughly an ellipse removing part of both ends of the long axis. According
each joint angle under the known position of the end effector, the wearable upper limb
to Figure 8 and kinematic analysis, it concludes that the working radius of the robot’s workspace
rehabilitation robot is solved of inverse kinematics.
on the X-axis and Y-axis is 795 mm, and the working radius on the Z-axis is 552 mm. According to
The inverse kinematics solution mainly includes two kinds of numerical solutions and closed
ergonomics, the average length of the medium human arm in China is 742 mm [38], indicating that
solutions. When the numerical method is used, the specific value of the joint variable can be
the limit position of the wearable upper limb rehabilitation robot is very close to the dynamic limit
obtained by using a recursive algorithm. The result of the numerical calculation can only be used.
position of the human upper limb. Therefore, the wearable upper limb rehabilitation robot designed to
The independent variable cannot be given at random and the calculated value can be obtained [39].
meet the needs of upper limb rehabilitation. Meanwhile, due to the limitation of mechanical structure
On the other hand, the closed solution method can be deduced according to the formula, and the
and its size, a cavity appears near the base of the wearable upper limb rehabilitation robot. The size of
dependent variable can be obtained by giving any independent variable. Because the closed
the cavity is relevant to the size of the base.
solution is more accurate and faster than the numerical method, and it is easy to distinguish all
possible
3.3. solutions,
Inverse the wearable upper limb rehabilitation robot designed in this paper satisfies the
Kinematics
Pieper criterion in robot kinematics, therefore, this paper uses the closed solution to solve, which
Wearable upper limb rehabilitation robots often assist patients with hemiplegia to complete some
provide a basis for subsequent structural optimization and motion control.
tasks in daily life, such as eating, holding things, and touching the head, to achieve the purpose of
Given the parameter values of D-H θ i can be obtained by inversely solving the following
restoring upper limb motor function. If the task is given, the end position is reached by the wearable
matrices:

T =10 T (θ1 )2 T (θ 2 )3 T (θ 3 )4 T (θ 4 )5 T (θ 5 )6 T (θ 6 )
0 1 2 3 4 5
6 (3)

Then,
Appl. Sci. 2020, 10, 2101 13 of 22

upper limb rehabilitation robot at this time. In order to determine the parameters of each joint angle
under the known position of the end effector, the wearable upper limb rehabilitation robot is solved of
inverse kinematics.
The inverse kinematics solution mainly includes two kinds of numerical solutions and closed
solutions. When the numerical method is used, the specific value of the joint variable can be obtained by
using a recursive algorithm. The result of the numerical calculation can only be used. The independent
variable cannot be given at random and the calculated value can be obtained [39]. On the other hand,
the closed solution method can be deduced according to the formula, and the dependent variable can
be obtained by giving any independent variable. Because the closed solution is more accurate and
faster than the numerical method, and it is easy to distinguish all possible solutions, the wearable
upper limb rehabilitation robot designed in this paper satisfies the Pieper criterion in robot kinematics,
therefore, this paper uses the closed solution to solve, which provide a basis for subsequent structural
optimization and motion control.
Given the parameter values of D-H θi can be obtained by inversely solving the following matrices:

0
6T =01 T (θ1 )12 T (θ2 )23 T (θ3 )34 T (θ4 )45 T (θ5 )56 T (θ6 ) (3)

Then,
−P y
!
−1
θ1 = tan + kπ (4)
Px
164 sin ϕ − 29 cos ϕ − b
!
−1
θ2 = tan + kπ (5)
a + 164 cos ϕ + 29 sin ϕ
x − y
θ3 = tan−1 + kπ (6)
z−u
!
s s5
θ4 = tan−1 4 + kπ (7)
c4 s5
 r33 
 c5 + c2 s3 + c3 s2 
−1 
θ5 = tan   + kπ (8)
(s2 s3 − c2 c3 )c4 

!
r32 v + r31 w
θ6 = tan−1 + kπ (9)
−r32 w + r31 v
where k means the integer value, and the value of k must be within the range of the value of θi at the joint
Py 1163−a2 −b2
 
angle; a = −5s1 − 110, b = P5z ; ϕ = sin−1 m − γ + kπ, m = q , γ = tan−1 328a +58b
58a−328b ;
(328a+58b)2 +(58a−328b)2
x = (170c2 − a)(164s2 − 29c2 ); y = (170s2 + b)(164c2 + 29s2 ); z = (170s2 + b)(29c2 − 164s2 );
u = (170c2 − a)(164c2 + 29s2 ), v = c2 s3 s5 + c3 s2 s5 − c2 c3 c4 c5 + c4 c5 s2 s3 .
The parameters of a certain number of position points which can be considered as the end of
the wearable upper limb rehabilitation robot are substituted into the above formula to obtain the
corresponding theoretical rotation angles of each joint. It is consistent with the rotation angle of each
joint in the 3D model. Thereby, it verifies the correctness of the inverse kinematics solution of the
wearable upper limb rehabilitation robot.
Since there are multiple inverse solutions in the process of solving inverse kinematics, this article
introduces three criteria for selecting solutions for robots to avoid singular solutions.
(1) Robot motion range requirements. During the structural design of the wearable upper limb
rehabilitation robot, the motion range of each joint is designed according to the requirements of the
upper limb motion range of the human body and to avoid mutual interference between mechanisms.
Therefore, the joints are required to meet the requirements of the joint motion range during the
movement, that is, the following inequality is satisfied:
Appl. Sci. 2020, 10, 2101 14 of 22

θmin
i ≤ θi ≤ θmax
i i = 1, 2, · · · , 6 (10)

Appl. Sci. 2020, 10, x FOR PEER REVIEW 15 of 23

where θmin
i
is the minimum limit angle of joint motion, and θmax i
is the maximum limit angle of
joint motion.
(2) Criterion of motion continuity. The application environment of the wearable upper limb
(2) Criterion of motion continuity. The application environment of the wearable upper limb
rehabilitation robot is to assist functional rehabilitation training for patients with motor dysfunction.
rehabilitation robot is to assist functional rehabilitation training for patients with motor dysfunction.
It requires the robot to continuously run smoothly and smoothly during the movement, no joint
It requires the robot to continuously run smoothly and smoothly during the movement, no joint angle
angle motion mutation, and to avoid the movement of the end effector of the robot. Therefore, in the
motion mutation, and to avoid the movement of the end effector of the robot. Therefore, in the actual
actual processing, the threshold value m is introduced to provide constraints and restrictions, that is,
processing, the threshold value m is introduced to provide constraints and restrictions, that is, to check
to check the absolute value of the difference between the next solution and the current joint angle
the absolute value of the difference between the next solution and the current joint angle value. If the
value. If the value is too large, the joint is abruptly changed, then another solution is selected.
value is too large, the joint is abruptly changed, then another solution is selected.
Δθ = θ n -θcur ≤ m
where θn ∆θof=joint
is the next position angle θcur | ≤ m
|θn −motion, and θcur is the current position angle(11)
of
joint motion.
where θn is the next position angle of joint motion, and θcur is the current position angle of joint motion.
(3)
(3) The
The principle
principle ofof minimum
minimum end end pose
pose error.
error. During
During thethe movement
movement of of the
the wearable
wearable upperupper limb
limb
rehabilitation robot, the position reached by the end effector and the target pose will inevitably have
rehabilitation robot, the position reached by the end effector and the target pose will inevitably have
errors. Therefore,
errors. Therefore,by bycomparing
comparing thethe final
final endendposepose transformation
transformation matrix,matrix,
the endthe end position
position and
and attitude
attitude error can be compared to achieve the goal of choosing the solution
error can be compared to achieve the goal of choosing the solution with the smallest error. with the smallest error.
During
During the theprocess
process of selection
of the the selection
for thefor the solution
inverse inverse ofsolution of theupper
the wearable wearable upper limb
limb rehabilitation
rehabilitation
robot, robot, the
the constrained constrained
conditions willconditions
also conflict willwith
alsoeach
conflict
other.with
At each other.the
this time, Atpriority
this time, the
of the
priority of the constraints should be set according to the actual situation, so
constraints should be set according to the actual situation, so as to obtain the optimal solution. as to obtain the optimal
solution.
4. Simulation Analysis of the Rehabilitation Robot Based on Virtual Prototype
4. Simulation Analysis of the Rehabilitation Robot Based on Virtual Prototype
4.1. Kinematic Simulation and Optimization Design
4.1. Kinematic Simulation and Optimization Design
In the structure of wearable upper limb rehabilitation robot, each joint is directly driven by motor.
In order to avoid
In the excessive
structure rigid impact
of wearable upperonlimb the affected limb, which
rehabilitation robot,causes secondary
each joint injury,driven
is directly ADAMS by
virtual prototype simulation software is exploited to simulate and analyze
motor. In order to avoid excessive rigid impact on the affected limb, which causes secondary injury, the motion stability of
wearable
ADAMS upper virtuallimb rehabilitation
prototype simulationrobot, and optimize
software the structure
is exploited accordingly.
to simulate The torque
and analyze of each
the motion
joint andof
stability the center ofupper
wearable mass at therehabilitation
limb end of the rehabilitation
robot, and robot
optimizewere selected
the structureas the measurement
accordingly. The
objects,
torque ofandeachthejoint
wholeandmotion timeofwas
the center mass setatatthe
5 s.end
Theofdriving motor rotated
the rehabilitation robotatwere
a constant
selectedspeed to
as the
study the motion
measurement stateand
objects, of the
theshoulder
whole motionjoint andtime elbow
was joint.
set at The
5 s. simulation
The drivingresults
motorare shown
rotated atina
Figures
constant9speed
and 10. to study the motion state of the shoulder joint and elbow joint. The simulation results
are shown in Figures 9 and 10.

Figure
Figure 9.
9. Torque transformation curves
Torque transformation curves of
of each
each joint.
joint.

Figure 10. End angular velocity/angular acceleration curve.

Appl. Sci. 2020, 10, 2101 15 of 22
Figure 9. Torque transformation curves of each joint.

Figure 10. End angular velocity/angular acceleration curve.

Specifically, Figure 9 shows the curve of joint torque changing with time when each joint is moving.
The initial state of the model is arm abduction to the state of maximum angle, and then adduction to the
position of maximum angle. Joint1, Joint2 and Joint3 are the torque variation curves when the shoulder
joint moves around three axes of three degrees of freedom simultaneously, and Joint4 is the torque
variation curve when the elbow joint moves at one degree of freedom. As can be observed, except
Joint1 of shoulder freedom, the torque of each joint fluctuates little at the start, then gradually flattens
to a certain value, and the motion curve is always smooth. It indicates that during the uniform motion
of each joint, the torque fluctuation of the joint is small and there is no big impact on the arm. Among
them, the torque of the elbow joint retains almost unchanged, indicating that the torque required by
the wrist joint is very small during the movement. This infers that during the uniform motion of each
joint, the torque fluctuation of the joint is small and there is no obviously impact on the arm.
Figure 10 plots the curves of angular velocity and angular acceleration changing with time when
the end is moving. The initial state of the model is arm abduction to the state of maximum angle, and
then adduction to the position of maximum angle. It can be seen from Figure 10 that the torque of each
joint fluctuates greatly at startup, and then the motion curve is smooth along with the movement of
each joint. This indicates that the end torque fluctuates less during the uniform motion of each joint,
and has no significantly impact on the hand.

4.2. Dynamic Simulation Analysis

In this paper, the dynamic analysis uses the three-dimensional model established in Section 2 into
Adams (Automatic dynamic analysis of mechanical systems) software, and then set the parameters of
the imported model, and add driving to each joint of the wearable upper limb rehabilitation robot.
In order to select the motor and ensure the output torque to satisfy the needs of rehabilitation, the
dynamic simulation of the wearable upper limb rehabilitation robot was carried out. Considering the
differences between patients with joint bearing capacity, for most people to bear upon initial value,
limb resistance to wearable upper limb rehabilitation robot is set to: 100 N in the horizontal movement
direction, 200 N in the vertical direction, the end points are in the hand close to the elbow, shoulder
rotation direction of 4 Nm torque, stress points in the forearm center of mass. The initial position of the
simulation is the wearable upper limb rehabilitation robot arm extending horizontally to the front,
and the simulation results are shown in Figure 11. The solid line in the diagram represents the torque,
while the dotted line indicates the rotation angle of the robot.
Appl. Sci. 2020, 10, 2101 16 of 22
Appl. Sci. 2020, 10, x FOR PEER REVIEW 17 of 23

Figure 11. Torques and angles of joints. (a) The rotation angle and the output torque of the drive
Figure 11. Torques and angles of joints. (a) The rotation angle and the output torque of the drive
motor when the shoulder joint moves in the vertical direction at different moments; (b) the relationship
motor when the shoulder joint moves in the vertical direction at different moments; (b) the
between the angle of rotation and the torque on the worm gear when the shoulder moves horizontally;
relationship between the angle of rotation and the torque on the worm gear when the shoulder
(c) the relationship between the rotation angle of the shoulder joint and the output torque of the motor;
moves horizontally; (c) the relationship between the rotation angle of the shoulder joint and the
(d) relationship between elbow flexion and extension angle and driving torque.
output torque of the motor; (d) relationship between elbow flexion and extension angle and driving
torque.
As shown in Figure 11, the dotted line is the joint rotation angle change curve, the variable is set
to θ, the solid line is the joint torque change curve T, and the variable is set to T = G · L(θ(t)). From
As shown in Figure 11, the dotted line is the joint rotation angle change curve, the variable is set
the torque formula L, where is the distance from the center of mass of the robot arm to the axis of the
to solid line is the joint torque change curve T , and the variable is set toT = G ⋅ L θ t .
θ , the joint.
shoulder ( ( ))
From the torque i−1 formula L , where is the distance from the center of mass of the robot arm to the
Ti = Ai = Rot(Z, θi ) × Trans(0, 0, di ) × Trans(ai , 0, 0) × Rot(X, ai )
axis of the shoulder joint.  
 cθi −sθi cαi sθi sαi ai cθi 
i −1
Ti = Ai = Rot (Zsθ, θi ) ×cθi cαi ( −cθi sα
Trans 0, 0, di )i × Trans
ai sθi ( ai , 0, 0 ) × Rot ( X , ai )

=  i
cθi −sθ i c0α i ssα θiisα i cα ai ci θi  di 


 0 0 0  1
=  sθi cθi cα i −cθi sα i ai sθi 
 0 sα i cα i di 
i−1 T is a transformation matrix between adjacent arm joints based on the D-H parameter model.
i
 
 0 0 0 L = 1 T_i [θ(t)]
X

i −1
Ti is a transformation matrix◦ between adjacent arm joints based on the D-H parameter
The maximum angle is θ = 120 in the initial position, and returns to the original position
model. ◦ ◦
θ = −280 , and then gos to the maximum angle θ = 120 .
=
More specifically, Figure 11a shows the relationship { _ [ between
( )]} the rotation angle and the output
torque of the drive motor when the shoulder joint moves in the vertical direction at different moments.
is θ 11a,
= 120
As can bemaximum
seen from angle
the Figure when0 the arm abducts to the maximum angle, the torque required is
The in the initial position, and returns to the original position
larger, and the mechanism is furthest from the static balance position. Figure 11b shows the relationship
θbetween
= −280 0
the angle of rotation
, and then gos toand the torque on
the maximum θ = 120
the worm
angle
0
gear .when the shoulder moves horizontally.
The wearable upper limb rehabilitation robot first swings up to the lowest point, then returns to the
More specifically, Figure 11a shows the relationship between the rotation angle and the output
initial point, and finally swings down to the maximum angle. The maximum torque of movement
torque of the drive motor when the shoulder joint moves in the vertical direction at different
in the horizontal direction occurs near the position from the top swing to the highest point, where
moments. As can be seen from the Figure 11a, when the arm abducts to the maximum angle, the
torque required is larger, and the mechanism is furthest from the static balance position. Figure 11b
Appl. Sci. 2020, 10, x FOR PEER REVIEW 18 of 23

shows the relationship between the angle of rotation and the torque on the worm gear when the
shoulder moves horizontally. The wearable upper limb rehabilitation robot first swings up to the
Appl. Sci.point,
lowest 2020, 10,then
2101 returns to the initial point, and finally swings down to the maximum angle. 17 The
of 22

maximum torque of movement in the horizontal direction occurs near the position from the top
swing to the highest
the required continuous point, where
torque theisrequired
time relatively continuous torque time
large, indicating is relatively
that the large, indicating
motor continues to work.
that
Figure 11c shows the relationship between the rotation angle of the shoulder joint and theangle
the motor continues to work. Figure 11c shows the relationship between the rotation outputof
the shoulder
torque of the joint
motor. andThetheforearm
output rotates
torque 600of the
in, motor.
600 outThe andforearm rotates
back to the 600 in,
starting 600 outThe
position. andmotor
back
to the starting position. The motor delivers torque through the gears
delivers torque through the gears and arc-shaped rack, providing a stable torque output for the elbow. and arc-shaped rack,
providing
Figure 11dashows stable the
torque output for
relationship the elbow.
between elbow Figure
flexion11dand shows the relationship
extension angle and between elbow
driving torque.
flexion
With theand extension
increase angle
of the angle,andthedriving
load oftorque.
the hand Withendthe increase
deviates of the
from theangle, the loadposition
static balance of the hand
and
end deviates from the static balance position and the torque increases.
the torque increases. As can be seen from Figure 11, through gear rack and other variable gear, As can be seen from Figure
the
11, through gear rack and other variable gear, the output torque of the
output torque of the drive motor is between a few Nm to dozens of Nm. The vertical direction of thedrive motor is between a few
Nm to dozens
shoulder joint of Nm. The
is larger, andvertical direction
the output torqueof ofthethe
shoulder joint is
direct drive larger, isand
method the output
generally in thetorque
tens ofof
the
N.mdirect drive method
to hundreds of Nm,iswhichgenerally in the
can be seentens N .m
thatofthe output to hundreds
performance of Nm, which can of
requirements bethe
seen that
motor
the
are output
greatly performance
reduced. According requirements of the motor
to the simulation are greatly
results, the DC reduced.
stepperAccording
servo motor to with
the simulation
planetary
results, the DC stepper servo motor with planetary reducer is
reducer is selected for the driving motor, which guarantees the accuracy of motion while selected for the driving motor, which
ensuring
guarantees
the power. the accuracy of motion while ensuring the power.
As
As the
the virtual
virtual prototype
prototype is is an ideal model,
an ideal model, and and thethe simplified
simplified driving
driving function
function is is adopted
adopted for for
joint rotation. During joint rotation, the PID controller in the actual machine
joint rotation. During joint rotation, the PID controller in the actual machine is not adopted when is not adopted whenthe
the rotation
rotation direction
direction is changed,
is changed, whichwhich
leadsleads
to theto local
the local
regionregion mutation,
mutation, which which
isn’tisn’t the focus
the focus of
of this
this paper, therefore, it is
paper, therefore, it is omitted here.omitted here.

5. Experimental
Experimental Setup
Setup and
and Results
Results
The wearable upper limb rehabilitation
rehabilitation robot
robot designed
designed in in this paper adopts the upper
computer-controlled robottotocarry
computer-controlled robot carryoutout continuous
continuous passive
passive rehabilitation
rehabilitation training
training foraffected
for the the affected
limb,
limb, and adopts
and adopts two rehabilitation
two rehabilitation trainingtraining
methods: methods: independent
independent movement movement of single
of single joint joint and
and combined
combined
functional functional
movementmovement
of multipleof multiple
joints. The joints. The block
block diagram of diagram of control
the overall the overall
systemcontrol system
is shown in
is shown
Figure 12.in Figure
The lower12. The lower
computer computer
of the control of the control
circuit takes thecircuit
motion takes the motion
control card andcontrol card and
data acquisition
data acquisition
card as card
the core. The as the core.
information The information
acquisition card is usedacquisition
to completecarddataiscollection
used toofcomplete data
3D gyroscope,
collection of 3D gyroscope,
motion acceleration and othermotion
sensors.acceleration
The upper and other can
computer sensors.
selectThe upper computer
the rehabilitation can select
training mode
the
andrehabilitation
set the motiontraining modeofand
parameters eachset the motion
joint, and then parameters
transmit the of related
each joint, and thentotransmit
parameters the
the motion
related parameters to the motion control card. The data collected by each functional
control card. The data collected by each functional sensor will be transmitted to the data acquisition sensor will be
transmitted to the datathe
card. After calculation, acquisition card. After
data acquisition card calculation,
converts thethe data acquisition
collected card converts
data into motion the
parameters,
collected data into
and then passes the motion control
parameters,
card toand
thethen
motor passes theinformation,
rotation motion control whichcard to the tomotor
is output each
rotation information,
motor driver which is output to each motor driver respectively.
respectively.

Figure 12. Block diagram of rehabilitation robot control system.

Appl. Sci. 2020, 10, 2101 18 of 22

To realize the precise control of wearable upper limb rehabilitation robots, and reduce the motor
torque error between the actual output torque and the theory, and let patients achieve the best effect
of rehabilitation, the torque demand larger shoulder joint is designed in this paper. In addition, the
drive motor of bending/stretching exercise and elbow flexion/stretch exercise drive motor installed on
the dynamic torque sensor based on the basis of the Section 4 results, that is to say, wearable upper
limbs rehabilitation robot dynamics analysis results of output torque. The dynamic torque sensor
transmits the information to the computer through the data acquisition card. The difference between
the theoretical torque and the measured torque is input into the PID controller and then the torque of
the motor is adjusted in real time to achieve the optimal control.
When designing a control system, it is essential to ensure the stability of the system and the safety
of patients. Some control methods and neural network technique have been paid more attention to
designing controller for a nonlinear system [40,41]. In order to prevent excessive output torque of
the motor from causing secondary damage to the affected limb, a current detection circuit is adopted
to feed the output current of the driver into the motion control card in real time. Once the detected
current exceeds the allowable value, the motor will stop immediately. At the same time, the one-key
emergency stop function is set so that the system can be stopped immediately when the patient is
uncomfortable. Furthermore, gyroscope and linear velocity sensor are used to detect the real-time
position, real-time speed and real-time acceleration of the motor, so as to avoid the adverse effect
caused by excessive joint rotation and excessive speed on the rehabilitation training.
Figure 13 depicts the completed wearable upper limb rehabilitation robot prototype system and
rehabilitation training. In the experiment, the time of each degree of freedom of the wearable upper
limb rehabilitation robot was set to 25 s, and the range of motion was from the initial position to the
extreme position. The test sampling period was 1 s. The experimental results are shown in Figure 14.
Specifically, the experiment curves of horizontal, vertical, rotation, and elbow flexion and extension of
the upper limb are plotted respectively. The solid red line in the figure is the simulation result, and the
blue dotted line is the test result. It can be observed, the actual angle and simulation angle of each joint
of the wearable upper limb rehabilitation robot during the movement have certain deviations. It is
mainly due to the error caused by the friction of the mechanical structure itself, and the subject wearing
the wearable upper limb rehabilitation robot will also be difficult to avoid rehabilitation. The wearable
upper limb rehabilitation robot applies a certain impedance, and one can see that the elbow joint
flexion and extension error is relatively large, because this part is driven by rope and the wire rope
needs to bear a large load, which increases the error caused by the elastic contraction of the wire rope
itself, but these errors do not affect the training characteristics of wearable upper limb rehabilitation
robot. The test results of the prototype test are consistent with the simulation results, revealing that the
wearable upper limb rehabilitation robot meets the expected design requirements and can complete
the motion functions required for passive rehabilitation training of the upper limbs.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 20 of 23

Appl. Sci.
Appl. Sci. 2020,
2020, 10,
10, 2101
x FOR PEER REVIEW 20 of
19 of 22
23
Appl. Sci. 2020, 10, x FOR PEER REVIEW 20 of 23

(a) (b)
Figure 13. Test and control
(a) system diagram of rehabilitation robot. (a) Tester (b)
wearing prototype test;
(b) physical map of control
(a) system. (b)
Figure 13. Test and control system diagram of rehabilitation robot. (a) Tester wearing prototype test;
(b) physical
Figure
Figure 13. map
13.Test
Test of control
and
and control system.
controlsystem
systemdiagram
diagramof
ofrehabilitation
rehabilitationrobot.
robot. (a)
(a) Tester
Testerwearing
wearingprototype
prototypetest;
test;
(b)physical
(b) physicalmap
mapofofcontrol
controlsystem.
system.

(a) (b)

(a) (b)
(a) 140 (b)
120
140
100
120
140
80
100
120
θ4 /(°)

60
80
100
40
θ4 /(°)θ4 /(°)

60 80
20
40 60
0 Simulation result
20 40 Test results
-20
00 5 10 15 20 25
20 Simulation result
t/sresults
Test
-200 Simulation result
0 5 10 15 20 25
Test results
-20 (d) t/s
(c) 0 5 10 15 20 25
t/s

Figure 14. Tracking results (d)

(c) of wearable upper limb rehabilitation robot. (a) Shoulder flexion/extension;
(b) shoulder abduction/adduction; (d) flexion/extension.
(c) shoulder internal/external rotation; (d) elbow
(c)
Appl. Sci. 2020, 10, 2101 20 of 22

6. Conclusions
Combined rehabilitation medicine and bionics, a 6-degree-of-freedom upper limb rehabilitation
robot was proposed, analyzed and investigated via using ropes as the mainstay and driving by the
“rope + toothed belt”. Combined with the human upper limb muscle anatomy characteristic and
relevant parameters, the mechanical mechanism of the wearable upper limb rehabilitation robot
was designed during the process of training movement. The workspace and kinematics analyses of
wearable upper limb rehabilitation robot were completed and analyzed, which laid the foundation
for the control scheme method of rehabilitation robot. The kinematics analysis of the shoulder joint
and joint motion was solely carried out in the horizontal and vertical directions, which resulted in
the relationship between the elbow joint flexion and extension angle and the driving torque. It also
provided a basis for the selection of the motor. The design of the wearable upper limb rehabilitation
robot control system was completed to ensure the safety of the affected limb during rehabilitation
training movement. The rehabilitation training experiments showed that the results were consistent
with the numerical simulation results, which could satisfy the requirements of rehabilitation training.
In addition, the movement transition between wearable upper limb rehabilitation robot and upper
limb was smooth, which might not lead secondary injury to the injured upper limb.
In future work, the structure optimization will be conducted to make the wearable upper limb
rehabilitation robot more compact and anthropomorphic. Furthermore, the control algorithm of
the wearable upper limb rehabilitation robot will be extensively investigated to enhance the control
precision and interaction ability of the robot system. Eventually, the advancement of the wearable
upper limb rehabilitation robot will in turn provide impetus for the development of rehabilitation
medical equipment in the context of upper limb rehabilitation training.

Author Contributions: In this work, Z.P., and T.W. conceived and designed the experiments; J.Y. gave some
constructive suggestions; Z.S. performed the experiments; Z.P., and S.L. analyzed the data; Z.W. guided the
writing of the article and made some modifications; and Z.P. wrote the paper. All authors have read and agreed to
the published version of the manuscript.
Funding: This work was supported in part by the National Natural Science Foundation of China under Grant
51875047 and Grant 61873304, in part by the China Postdoctoral Science Foundation funded project under Grant
2019T120240 and 2018M641784, and in part by the Foundation of Jilin Province Science and Technology under
Grant 20170307012YY.
Acknowledgments: The authors are grateful to the anonymous reviewers and the Editor for their valuable
comments and suggestions on improving this paper.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Yang, Q.Z.; Cao, D.F.; Zhao, J.H. Analysis on state of the art of upper limb rehabilitation robots. Robot 2013,
35, 630–640. [CrossRef]
2. Marichal, S.; Malaisé, A.; Modugno, V.; Dermy, O.; Charpillet, F.; Ivaldi, S. One-shot evaluation of the control
interface of a robotic arm by non-experts. In Proceedings of the 8th International Conference on Social
Robotics, Kansas City, MO, USA, 1–3 November 2016; Volume 9979, pp. 458–468. [CrossRef]
3. Wu, Q.C.; Wang, X.S.; Chen, B.; Wu, H.T. Development of an RBFN-based neural-fuzzy adaptive control
strategy for an upper limb rehabilitation exoskeleton. Mechatronics 2018, 53, 85–94. [CrossRef]
4. Lo, H.S.; Xie, S.Q. Exoskeleton robots for upper-limb rehabilitation: State of the art and future prospects.
Med. Eng. Phys. 2012, 34, 261–268. [CrossRef]
5. Li, H.R.; Jiang, Z.B.; Wu, W. Kane dynamic equations based on screw theory for upper limb rehabilitation
robot. J. Northeast. Univ. 2014, 35, 153–157. [CrossRef]
6. Niyetkaliyev, A.S.; Hussain, S.; Ghayesh, M.H.; Alici, G. Review on design and control aspects of robotic
shoulder rehabilitation orthoses. IEEE Trans. Hum. Mach. Syst. 2017, 47, 1134–1145. [CrossRef]
7. Hu, W.L.; Li, G.F.; Sun, Y.; Jiang, G.Z.; Kong, J.Y.; Ju, Z.J.; Jiang, D. A review of upper and lower limb
rehabilitation training robot. In Proceedings of the 2017 International Conference on Intelligent Robotics and
Applications, Wuhan, China, 16–18 August 2017; Volume 10462, pp. 570–580. [CrossRef]
Appl. Sci. 2020, 10, 2101 21 of 22

8. Perry, J.C.; Rosen, J.; Burns, S. Upper-limb powered exoskeleton design. IEEE/ASME Trans. Mechatron. 2007,
12, 408–417. [CrossRef]
9. Marco, G.; Alexander, D.W.; Simon, B.; Verena, K.M.; Tobias, N.; Rober, R. A robotic system to train activities
of daily living in a virtual environment. Med. Biol Eng. 2011, 49, 1213–1223. [CrossRef]
10. Colomer, C.; Baldoví, A.; Torromé, S.; Navarro, M.D.; Moliner, B.; Ferri, J.; Noé, E. Efficacy of Armeo® Spring
during the chronic phase of stroke. Study in mild to moderate cases of hemiparesis. Neurología 2013, 28,
261–267. [CrossRef]
11. Sugar, T.G.; He, J.H.; Koeneman, E.J.; Koeneman, J.B.; Herman, R.; Huang, H.; Schultz, R.S.; Herring, D.E.;
Wanberg, J.; Balasubramanian, S.; et al. Design and control of RUPERT: A device for robotic upper extremity
repetitive therapy. IEEE Trans. Neural Syst. Rehabil. Eng. 2007, 15, 336–346. [CrossRef]
12. Otten, A.; Voort, C.; Stienen, A.; Aarts, R. LIMPACT: A hydraulically powered self-aligning upper limb
exoskeleton. IEEE/ASME Trans. Mechatron. 2015, 20, 2285–2298. [CrossRef]
13. Wang, X.; Song, Q.Z.; Wang, X.G.; Liu, P.Z. Kinematics and dynamics analysis of a 3-DOF upper-limb
exoskeleton with an internally rotated elbow joint. Appl. Sci. 2018, 8, 464. [CrossRef]
14. Nam, H.S.; Hong, N.Y.; Cho, M.W.; Lee, C.W.; Seo, H.G.; Kim, S.W. Vision-assisted interactive
human-in-the-loop distal upper limb rehabilitation robot and its clinical usability test. Appl. Sci. 2019, 9,
3106. [CrossRef]
15. Tang, Z.C.; Zhang, K.J.; Sun, S.Q.; Gao, Z.G.; Zhang, L.K.; Yang, Z.L. An upper-limb power-assist exoskeleton
using proportional myoelectric control. Sensors 2014, 14, 6677–6694. [CrossRef] [PubMed]
16. Bai, J.; Song, A.G.; Li, H.J. Design and analysis of cloud upper limb rehabilitation system based on motion
tracking for post-stroke patients. Appl. Sci. 2019, 9, 1620. [CrossRef]
17. Li, Z.J.; Su, C.Y.; Wang, L.Y.; Chen, Z.T.; Chai, T.Y. Nonlinear disturbance observer-based control design for
a robotic exoskeleton incorporating fuzzy approximation. IEEE Trans. Ind. Electron. 2015, 62, 5763–5775.
[CrossRef]
18. Huang, J.; Huo, W.G.; Xu, W.X.; Samer, M.; Yacine, A. Control of upper-limb power-assist exoskeleton using
a human-robot interface based on motion intention recognition. IEEE Trans. Autom. Sci. Eng. 2015, 12,
1257–1270. [CrossRef]
19. Mao, Y.; Agrawal, S.K. Design of a cable-driven arm exoskeleton (CARES) for neural rehabilitation. IEEE
Trans. Robot. 2012, 28, 922–931. [CrossRef]
20. Wu, Q.C.; Wang, X.S.; Wu, H.T.; Chen, B. Research on the gravity balance characteristics of an upper limb
rehabilitation exoskeleton. Robot 2017, 39, 81–88. [CrossRef]
21. Wu, Q.C.; Wang, X.S.; Du, F.P.; Zhu, Q. Fuzzy sliding mode control of an upper limb exoskeleton for
robot-assisted rehabilitation. In Proceedings of the 2015 IEEE International Symposium on Medical
Measurements and Applications, Turin, Italy, 7–9 May 2015; pp. 451–456. [CrossRef]
22. Wu, Q.C.; Wang, X.S.; Du, F.P.; Xu, J.G. Development and control of a Bowden-cable actuated exoskeleton for
upper-limb rehabilitation. In Proceedings of the 2014 IEEE International Symposium on Robotic and Sensors
Environments, Timisoara, Romania, 16–18 October 2014; pp. 7–12. [CrossRef]
23. Frisoli, A.; Borelli, L.; Montagner, A.; Marcheschi, S.; Procopio, C.; Salsedo, F.; Bergamasco, M.;
Carboncinit, M.C.; Tolainit, M.; Rossit, B. Arm rehabilitation with a robotic exoskeleleton in Virtual
Reality. In Proceedings of the IEEE 10th International Conference on Rehabilitation Robotics, Noordwijk,
The Netherlands, 12–15 June 2007; pp. 631–642. [CrossRef]
24. Cortese, M.; Cempini, M.; Ribeiro, P.R.; Soekadar, S.R.; Carrozza, M.C.; Vitiello, N. A Mechatronic System for
Robot-Mediated Hand Telerehabilitation. IEEE/ASME Trans. Mechatron. 2015, 20, 1753–1764. [CrossRef]
25. Roh, J.; Rymer, W.Z.; Perreault, E.J.; Yoo, S.B.; Beer, R.F. Alterations in upper limb muscle synergy structure
in chronic stroke survivors. J. Neurophysiol. 2013, 109, 768–781. [CrossRef]
26. Frisoli, A.; Edoardo, S.; Procopio, C.; Bergamasco, M.; Chisari, C.; Lamola, G.; Rossi, B. Training and
assessment of upper limb motor function with a robotic exoskeleton after stroke. In Proceedings of the 2012
4th IEEE International Conference on Biomedical Robotics and Biomechatronic, Roma, Italy, 24–27 June 2012;
pp. 1782–1787. [CrossRef]
27. Laribi, M.A.; Decatoire, A.; Carbone, G.; Pisla, D.; Zeghloul, S. Identification of upper limb motion
specifications via visual tracking for robot assisted exercising. In Proceedings of the International Conference
on Robotics in Alpe-Adria-Danube Region, Patras, Greece, 29 September 2018; pp. 93–101. [CrossRef]
Appl. Sci. 2020, 10, 2101 22 of 22

28. Wang, B.H.; Chen, S.; Zhu, B.; Liang, Z.W.; Gao, X. Design and Implementation of Shoulder Exoskeleton
Robot. Int. Conf. Soc. Robot. 2018, 11357, 241–252. [CrossRef]
29. Major, K.A.; Major, Z.Z.; Carbone, G.; Pisla, A.; Vaida, C.; Gherman, B.; Pisla, D.L. Ranges of motion as basis
for robot-assisted post-stroke rehabilitation. Hum. Vet. Med. 2016, 8, 192–196.
30. Neumann, D.A. Kinesiology of the Musculoskeletal System: Foundations for Rehabilitation, 3th ed.; Mosby: St.
Louis, MO, USA, 2016.
31. Rahman, M.; Thierry, K.O.; Saad, M.; Kenne, J.P. Robot assisted rehabilitation for elbow and forearm
movements. Int. J. BioMechatron. Biomed. Robot. 2011, 1, 206–218. [CrossRef]
32. Abdallah, I.B.; Bouteraa, Y.; Rekik, C. Web-based robot control for wrist telerehabilitation. In Proceedings
of the 2016 4th International Conference on Control Engineering & Information Technology, Hammamet,
Tunisia, 16–18 December 2016. [CrossRef]
33. Gates, D.H.; Walters, L.S.; Cowley, J.; Wilken, J.M.; Resnik, L. Range of Motion Requirements for Upper-Limb
Activities of Daily Living. Am. J. Occup. Ther. 2016, 70, 1–10. [CrossRef] [PubMed]
34. Chen, Y.Y.; Li, G.; Zhu, Y.H.; Zhao, J.; Cai, H.G. Design of a 6-DOF upper limb rehabilitation exoskeleton
with parallel actuated joints. BioMed. Mater. Eng. 2014, 24, 2527–2535. [CrossRef]
35. Trigili, E.; Grazi, L.; Crea, S.; Accogli, A.; Carpaneto, J.; Micera, S.; Vitiello, N.; Panarese, A. Detection of
movement onset using EMG signals for upper-limb exoskeletons in reaching tasks. J. Neuroeng. Rehabil. 2019,
16, 45. [CrossRef]
36. Pang, Z.X.; Zhang, B.C.; Yu, J.Z.; Sun, Z.B.; Gong, L.N. Design and analysis of a Chinese medicine based
humanoid robotic arm massage system. Appl. Sci. 2019, 9, 4294. [CrossRef]
37. Jin, X.J.; Jun, D.I.; Jin, X.M.; Park, S.K.; Park, J.O.; Ko, S.Y. Workspace analysis of upper limb for a planar
cable-driven parallel robots toward upper limb rehabilitation. In Proceedings of the 2014 14th International
Conference on Control, Automation and Systems, Seoul, Korea, 22–25 October 2014; pp. 352–356. [CrossRef]
38. Shi, Y. Human Factors Engineering; Tsinghua University Press: Beijing, China; Beijing Jiaotong University
Press: Beijing, China, 2011; pp. 19–31.
39. Wang, Z.M.; Chang, Y.Y.; Sui, X.T. Kinematics analysis and trajectory planning of upper limb rehabilitation
robot. In Proceedings of the 2017 International Conference on Industrial Informatics-Computing Technology,
Intelligent Technology, Industrial Information Integration, Wuhan, China, 2–3 December 2017; pp. 177–180.
[CrossRef]
40. Sun, Z.B.; Li, F.; Zhang, B.C.; Sun, Y.Y.; Jin, L. Different modified zeroing neural dynamics with inherent
tolerance to noises for time-varying reciprocal problems: A control-theoretic approach. Neurocomputing 2019,
337, 165–179. [CrossRef]
41. Sun, Z.B.; Tian, Y.T.; Wang, J. A novel projected Fletcher-Reeves conjugate gradient approach for finite-time
optimal robust controller of linear constraints optimization problem: Application to bipedal walking robots.
Optim. Control Appl. Methods 2018, 39, 130–159. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).