See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224089368

Fuzzy PID control of a wearable rehabilitation robotic hand driven by

pneumatic muscles

Conference Paper · December 2009

DOI: 10.1109/MHS.2009.5352012 · Source: IEEE Xplore

CITATIONS READS

35 856

5 authors, including:

Jun Wu Jian Huang

Huawei Technologies Co., Ltd Huazhong University of Science and Technology
23 PUBLICATIONS 464 CITATIONS 346 PUBLICATIONS 6,769 CITATIONS

SEE PROFILE SEE PROFILE

Yongji Wang Kexin Xing

Chinese Academy of Sciences Zhejiang University of Technology
282 PUBLICATIONS 3,038 CITATIONS 17 PUBLICATIONS 631 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Jun Wu on 07 June 2015.

The user has requested enhancement of the downloaded file.

 Fuzzy PID Control of A Wearable Rehabilitation Robotic Hand Driven by
Pneumatic Muscles
Jun Wu, Jian Huang, Member,IEEE, Yongji Wang, Kexin Xing, Qi Xu

Key Laboratory of Image Processing and Intelligent Control, Department of Control Science & Engineering,
Huazhong University of Science and Technology, 430074, Wuhan, China

Abstract: assist hand therapy.

Intensive task-oriented repetitive physical therapies need be Many kinds of assist devices for hand therapy have been
provided by individualized interaction between the patients developed to offer many patients the intensive training that is
and the rehabilitation specialists to improve hand motor needed for neural reorganization and functional changes. Some
performance for those survived from stoke and traumatic exoskeleton based systems are too heavy and bulky, or too
brain injury. The goal of this research is to develop a novel expensive for home use by individuals with hand impairments.
wearable device for robotic assisted hand repetitive therapy. Recently, the compliant pneumatic muscle actuators (PMA) have
We designed a pneumatic muscle (PM) driven therapeutic found many applications in the robot-based hand motor therapy. A
device that is wearable and provides assistive forces required wearable power assist device[3] for hand grasping is proposed to
for grasping and finger extension. The robot has two distinct improve the quality of life (QOL) of the elderly and disabled
degrees of freedom at the thumb and all other fingers. The
persons. The hand Mentor™[4] is designed to offer active
embedded sensors can feedback position and force information
repetitive therapy to the fingers and the wrist by encouraging
for robot control and quantitative evaluation of task
performance. It is potential of providing supplemental at-home self-powered motion and assisting movement when necessary. The
therapy in addition to in the clinic treatment. To realize the hand wrist assistive rehabilitation device with 3 degrees of freedom
trajectory tracking control, a fuzzy PID controller is designed is developed to assist the hand in grasp and release movements[5].
for the proposed device. The experimental results show angle Compared with the electric motor technology, the PMA is soft and
tracking control of the robotic hand using the fuzzy PID exhibits many properties of human muscle. Therefore, the hand
controller has better performance than that using a therapy device with PMAs can interact with the patient in a safer
conventional PID controller. and more natural way[6]. So we designed a novel wearable device
for robotic assisted hand repetitive therapy by PMA.
1. INTRODUCTION In the assisted robotics literature, PM is at a very low level of
risk of injury. This is not the case for hydraulic and/or electric
Stroke is a leading cause of disability, commonly involving motor actuators, which produce a far greater risk of injure to the
deficits of motor functions. The findings that repetitive motor human user adjacent to the device when failing. Pneumatic muscle
function activities can induce neural plasticity and result in is an interesting actuator, very similar in action to human skeletal
functional recovery form the basis for new therapeutic treatment of muscle, which has several important advantages.
patients who survived stroke and traumatic brain injury. However, (1) Simple structure, light weight, ease of miniaturization;
traditional approach and technique of repetitive function training (2) Large output force-self-weight rate, high energy conversion
for neurological rehabilitation are very labor intensive and lack of efficiency, small noise;
consistency and objective assessment. There has been a continuous (3) Low cost, easy maintenance;
effort by engineers to develop robotic systems that can assist and (4) Perform like a "soft actuator".
improve the rehabilitation outcomes of patients with The main problem of pneumatic muscle actuators is that this
neuromuscular disabilities. Many robotic systems tend to be kind of actuators are highly nonlinear and uncertainty systems.
expensive and are often developed for research purpose, therefore, Lack of a common description of the mathematical model,
too complex for practical use. There is a need for low cost, safe and traditional methods of control based on the mathematical model is
practical devices to assist therapy [1]-[6]. difficult to achieve. The nonlinearity makes the design of a
Hand is an organ of grasp as well as sensation, fine controller more difficult.
discrimination and exquisite dexterity. A normally functioning It is well known that up until now, conventional PID controller
hand is very important for an independent and active life. is most widely used in industry due to its simple control structure,
Unfortunately, the incidence of injuries directly affecting the hand easy implementation, and the model of controlled plant is not
has risen dramatically in the recent years[7]. These injuries include required[7][8]. But its parameters are not adaptive. Some optimal
traumatic injuries, congenital deformities, neurological and control methods have more precision. Various types of modified
arthritic conditions, and regional pain syndrome of the hands. Due conventional PID controllers such as auto-tuning and adaptive PID
to the central role that hand movements normally play in human controllers were developed lately. Also, a class of nonconventional
existence, much attention in rehabilitation research has been type of PID controller employing fuzzy logic has been designed
focused on understanding and restoring hand motor function after and simulated for this purpose.
injury. There is a need for low cost, safe and practical devices to Stability of these fuzzy PID controllers has also been analyzed

     © )%%% - 408 -

Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on May 30,2010 at 14:11:06 UTC from IEEE Xplore. Restrictions apply.
 and is guaranteed. Many simulation and practical examples have which made in lab are about 25%. In order to magnify the limited
been given to show the superior performance of fuzzy PID travel of the PMAs, a pulley mechanism is designed and arranged
controllers[9]-[11]. In this paper, a self-tuning parameter angle above the metacarpophalangeal joints.
controller based on fuzzy PID for the wearable rehabilitation
robotic hand which is proposed. The parameters of PID can be
tuned online according to the different error (which is between the
real angle and expectation) and error variety rate. It not only can
keep the advantages of the classical PID controller, but also
possesses higher agility and adaptability.
The organization of this paper is as follows: the next section
mainly describes the mechanical construction. In section III the
control structure and control algorithm are addressed. The
experiment is presented in section IV. In the last section, the
conclusion of the current work and the possible future Fig.2. Materials of pneumatic muscle actuator made in lab
improvements are described.

2. WEARABLE REHABILITATION ROBOT HAND

The design goal is to develop a hand therapeutic device that

provides training of grasping and finger extending motions for
activities of daily living. The device actively involves the patient in
their hand rehabilitation by encouraging self-powered motion by
the fingers and thumb, and assisting movement when necessary.
The device is to be compact for portability, relatively easy to use
Fig.3. Pneumatic muscle actuator (Mckibben Muscle) made in lab
for self don and doff, capable of interacting with a personal
computer based visual biofeedback system to capture the interest of
the user, provides measurement of functional performance and has
the potential to be inexpensive and amenable for home and clinical
use.

2.1 Mechanic Structure

Traditional robots are usually stiff. The use of robotic actuation
in exoskeleton applications presents a mismatch in compliance of
the actuator and the limb being moved. The "Mckibben" type
pneumatic actuators used in the proposed device provide compliant Fig.4. Prototype of wearable device for hand rehabilitation
actuation and thus augment the safety compared to stiffer drives. Fig.4 shows the prototype of wearable rehabilitation device for
Fig.1 shows the structure of developed device for hand the left hand. The majority of the lightweight device is
rehabilitation. This device consists of a PMA (pneumatic muscle manufactured by the thermoplastic plastics. The foam-lined
actuator) put in the back of finger and a PMA put in the back of sub-plate that is attached to the forearm and hand with velcro straps
thumb. provides a comfortable wear for the patients. Most patients such as
the stroke survivors have the ability to flex the joints of the thumb
and fingers due to the familiar spasticity, so the flexion function
needed for hand grasping is left to the patient. The preliminary
hand therapy device provides active assistance at two degrees of
freedom for release movements: finger extension and thumb
extension. The new device incorporates coordinated thumb and
finger motions that are the required finger movement in daily life.
Furthermore, all assisted movements will be made in the same slow
manner as is done with the current wrist/finger device to protect
against significant increases in muscle tone.
Fig.1. Structure of wearable device for hand rehabilitation. 1-force sensors,
2-pneumatic muscle, 3-forearm attachment, 4-angle sensor, 5-hand
attachment, 6-thumb cup, 7-travel transform pulley, 8-glide mechanism, 2.2 Fundamental Motion
9-angle sensor, 10-press plate, 11-pulley, 12-finger splint, 13-angle sensor, The extension of the thumb and fingers in able-bodied
14-lever volunteers with the wearable robotic device are done as the
PMA is made from inexpensive materials, such as natural latex fundamental experiment. The robot is put on the left human hand,
rubber for the inner bladder covered by a polymer based braid then the angles of joints of the thumb and fingers as shown in Fig.5
(Fig.2 and Fig.3). The maximum contraction ratios of both PMAs are measured when the robotic device works.

- 409 -

Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on May 30,2010 at 14:11:06 UTC from IEEE Xplore. Restrictions apply.
 For a wearable therapy device, the safety is dominant within all
performances. The affected hand needs to be absolutely safe when
interacting with the robotic device. Firstly, the risk to patients is
minimized in this design by utilizing the compliant PMAs. All
assisted movements will be made slow to protect against
significant increases in muscle tone. Secondly, the mechanical
design allows the device physically to limit range of motion of
individual joints. Force application is limited by the compliant
nature of the actuators. Moreover, the safety envelope in the
Fig.5. Disposition of joints control algorithm will further guarantee the safety. If any of the
PMA forces exceeds the preset limit, the robot device is shut down.
Table 1 shows angles of joint where ș0 means the initial angle, In addition, a panic button is provided for easy access by the
įș represents the increased angle and ș is the current angle. ș0 , įș unaffected hand to exhaust the PMAs and shut the device down if
and ș are illustrated respectively in Fig.6 and Fig.7. the patient is feeling any discomfort, fatigue or anxiety.
TABLE 1. THE RANGE OF MOTION FOR ROBOT
PIP[rad/deg.] MP[rad/deg.] CM[rad/deg.] 3. CONTROL SYSTEM
ș0 1.6057/92 1.5708/90 0.3840/22
3.1 Control Structure
įș 1.5359/88 1.5708/90 1.1868/68 Task-oriented repetitive movements can improve muscular
strength and movement coordination in patients with impairments
ș 3.1416/180 3.1416/180 1.5708/90
due to neurological problems. A typical repetitive hand functional
movement is grasping (flexion) and release (extension). This study
deals with the application of human-centered robotic approaches to
robot-aided hand rehabilitation of neurological disorders. During
movement the robot control system takes into account the patient’s
intention and voluntary efforts rather than imposing any predefined
movements. The behavior of robotic device adapts to the existing
voluntary motor abilities, displays and improves the patient
contribution by visual biofeedback.
Fig.7 shows the block diagram of control system for the robot.
Fig.6. Initial angle and increased joint angle for the fingers and thumb.
ș is the joint angle. ș0 means the initial angle, and įș represents the
The goal of control scheme is to provide controllable, quantifiable
increased angle. assistance specific to a particular patient by adapting the level of
the assistance provided.
It will allow patient to achieve a movement objective with the
2.3 Safety

Fig.7. Block diagram of control system for the robotic hand rehabilitation device

- 410 -

Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on May 30,2010 at 14:11:06 UTC from IEEE Xplore. Restrictions apply.
 least assistive force. The sensors can feedback position and force The self-tuning fuzzy PID controller is the integration of a fuzzy
information for adaptation of the assistive force and quantitative controller and a conventional PID controller. Considering the fuzzy
evaluation of task performance. The angle sensors are arranged in relationship between PID parameters and the control deviation E
the metacarpophalangeal joints and interphalangeal joints of thumb and deviation change rate Ec, the fuzzy logic rules are designed. In
and fingers to measure the joint flexion and extension. The force the control process, the fuzzy controller calculates coefficients kp,
sensors are used to measure the PMA forces of fingers and thumb. ki and kd online and then adjusts the PID parameters. The fuzzy
In addition, there is also a pressure sensor integrated in the electric rules are shown in Table 2, 3, 4.
proportional valve to provide the local feedback. Based on the The basic domain, fuzzy domain and quantification factors are
quantitative estimation of the patient’s active muscular effort and shown in Table 5 are used for the adjustment of fuzzy PID
spasticity in terms of the moments and the movement produced in parameters. We choose triangular shape membership functions for
the finger and thumb joint. The control algorithms provide optimal all variables to implement fuzzification.
assistance to each patient, which can be defined as the least amount
or "just enough" assistance that is necessary to enable the patient to TABLE 2. KP’S FUZZY REGULATION RULES
make a specific movement. E
Ec
The real-time data acquisition card (Advantach USB-4716) is NB NM NS ZO PS PM PB
responsible for sampling all sensory data. The analog output of the NB PB PB PM PM PS ZO ZO
card is connected to the electric proportional valve to control the NM PB PB PM PS PS ZO PS
PMA. The air transmission unit is an air-conditioning circuit which NS PM PM PM PS ZO PS PS
contains filter valve and electric proportional valve. The data is ZO PM PM PS ZO PS PM PM
exchanged via a USB connection between the data acquisition card PS PS PS ZO PS PS PM PM
and host computer. The host computer runs the control algorithms PM PS ZO PS PM PM PM PB
and virtual reality game. It can build a task-oriented interactive PB ZO ZO PM PM PM PB PB
game environment to deliver an engaging and interesting training
process. The evaluation of task performance is fed back to patients TABLE 3. KI’S FUZZY REGULATION RULES
by intuitive cues integrated within the computer games. E
Ec
NB NM NS ZO PS PM PB
3.2 Self-tuning fuzzy PID Controller
The efficient and stable control algorithm is the key for the NB ZO PS PB PB PB PS ZO
angle tracking smoothly and quickly. According to the demand of NM ZO PS PB PB PB PS ZO
angle control, fuzzy PID algorithm is used. This paper combines NS ZO ZO PM PB PM ZO ZO
fuzzy control with PID control, using the idea of fuzzy reasoning, ZO ZO ZO PM PB PM ZO ZO
according to the control deviation e and deviation change rate ǻe to PS ZO ZO PM PB PM ZO ZO
tune the parameters kp, ki and kd online. The controller can meet PM ZO PS PB PB PB PS ZO
the desire of self-tuning parameters based on the error e and the
PB ZO PS PB PB PB PS ZO
rate ǻe of change of error as its inputs. The structure diagram of
control system shows in Fig. 8. The incremental PID Control TABLE 4. KD’S FUZZY REGULATION RULES
algorithm is shown as Eqn. (1). E
'u (k ) k p [ e ( k )  e ( k  1)]  k i e ( k ) Ec
(1) NB NM NS ZO PS PM PB
 k d [ e ( k )  2 e ( k  1)  e ( k  2 )] NB PS PM ZO ZO ZO PM PS
where ǻu(k) is the control increment at k-th sampling time, and e(k) NM PS PM PS ZO PS PM PS
is the deviation of the system at the k-th sampling time. kp is the NS PB PB PS ZO PS PB PB
proportional gain. ki = kp*T/Ti is the integral coefficient. kd = ZO PB PB PS ZO PS PB PB
kp*Td/T is the derivative coefficient. In the process of system PS PB PB PS ZO PS PB PB
debugging, the initial value of kp, ki and kd were determined by PM PS PM PS ZO PS PM PS
conventional methods. PB PS PM ZO ZO ZO PM PS

TABLE 5. FUZZY PID PARAMETERS TABLE

Variables e(º) ǻe(º/s) kp ki kd
Linguistic
E Ec Kp Ki Kd
variables
Basic [-0.9 [-9.0 [0 [0 [0
domain 0.9] 9.0] 0.12] 0.005] 0.002]
Fuzzy
[-6 6] [-6 6] [0 1] [0 1] [0 1]
domain
Quantificat-
Fig.8. Control System Structure Diagram 20/3 2/3 3/25 1/200 1/500
ion factors

- 411 -

Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on May 30,2010 at 14:11:06 UTC from IEEE Xplore. Restrictions apply.
 control loops run independently for the two pneumatic muscles
4. EXPERIMENT RESULTS which drive the four fingers and the thumb to move. The controlled
variable for each control loop is the measured angle at each joint.
4.1 Experimental Condition The software running on the PC was developed using Visual C++
Fig.4 and Fig.7 show the experimental apparatus and schematic 6.0.
diagram for the control of the wearable device which is used in TABLE.6 THE PARAMETERS OF PNEUMATIC MUSCLE FOR MP
hand rehabilitation training.the experimental apparatus including: Initial length L0 225mm
pneumatic muscles, the electromagnetic proportional valve, Initial diameter D0 12.26mm
pressure regulating valve, displacement sensors, angle sensors and Initial angle of mesh grid ș0 22º
the Advantech’s acquisition cards. Table.6-8 show the relative Thickness of rubber sleeve tk 1.64mm
parameters of pneumatic muscles and equipments. TABLE.7 THE PARAMETERS OF PNEUMATIC MUSCLE FOR CM
The parameters of conventional PID controller are: kp=0.08, ki Initial length L0 204mm
=0.003 and kd =0.0015. In a fuzzy PID way, kp, ki and kd can be Initial diameter D0 12.26mm
calculated according to Table2-5 and changed within certain range. Initial angle of mesh grid ș0 22º
The control algorithms were implemented in the computer. Two Thickness of rubber sleeve tk 1.64mm
TABLE.8. THE MAIN PERFORMANCE PARAMETERS OF THE EQUIPMENTS
Name Model Performance Manufacturer
Mute air mpressor FB-0.017/7 Rating exhaust pressure: 0.7Mpa Taiwan JAGUAR Co., Ltd
ITV1030-211BS Input: 0-5V
Electromagnetic proportion valve SMC
Output: 0.005-0.5Mpa
GA-75 Range: 0-150mm Beijing Jinghaiquan Sensing
Displacement sensor
Linearity: <0.1%(=0.09%) Technological Co., Ltd
Pressure regulating valve AW20-02BCG Range: 0.05-0.85Mpa SMC Co., Ltd
USB-4716 8 channel DI, 8 channel DO
Data acquisition card 16 channel AI, 2 channel AO Advantech Co., Ltd
sampling rate: 200 kS/s
Q-120DE Input: AC 220V±10% 50Hz
Switching power supply Taiwan MEAN WELL
Output: ±5V, ±12V, ±24V
RDC50 Degree range: 320
Angle sensor ALPS electric CO., LTD
linearity: ±2%
Others Joints connector, air pipe, etc SMC Co., Ltd

4.2 Result were evaluated using ramp wave as the goal of joint angle
In order to further assess the performance of the system and the trajectory tracking for the extensions of the MP joint and CM joint.
realization of the ultimate clinical application, we conducted a The user was passive and not working with or against the motion.
preliminary clinical trials, as shown in Figure 9. The angle tracking result of MP using the conventional PID
and the fuzzy PID control strategies is showed in Fig.10. The angle
tracking result of CM using the two control strategies is showed in
Fig.11.
3.1
PID
Fuzzy PID
3

2.9
(a) (b)
Joint angle(rad)

2.8

2.7

2.6

2.5

(c) (d) 2.4

0 0.5 1 1.5 2 2.5 3 3.5 4
Fig.9. The sequence of a clinical rehabilitation trial Time(s)

Some experiments were performed on a healthy subject. In the Fig.10. The result of metacarpophalangeal joint (MP) tracking with PID
first experiment, the performance of PID and fuzzy PID controllers and adaptive fuzzy PID control algorithms

- 412 -

Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on May 30,2010 at 14:11:06 UTC from IEEE Xplore. Restrictions apply.
 hand impairment back into the community.
1.6
PID
Fuzzy PID ACKNOWLEDGMENT
1.5

This work is partially supported by Hi-tech Research and

1.4
Development Program of China under Grant 2007AA04Z204 and
Grant 2008AA04Z207, and in part by the Natural Science
Joint angle(rad)

1.3 Foundation of China under Grant 60674105.

1.2
REFERENCES
1.1
[1] J. He, E. J. Koeneman, R. S. Schultz, D. E. Herring, J.
Wanberg, and H. Huang et al., “RUPERT: A device for
1 robotic upper extremity repetitive therapy,” in Proceedings of
the 27th Annual Conference of the IEEE Engineering in
0.9
0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2
Medicine and Biology, Shanghai, China, 2005, pp.6844-6847.
Time(s) [2] I. Sarakoglou, N. G. Tsagarakis and D. G. Caldwell,
“Occupational and physical therapy using a hand exoskeleton
Fig.11. The result of thumb carpometacarpal joint (CM) tracking with PID
and adaptive fuzzy PID control algorithms based exerciser,” in Proceedings of 2004 IEEE/RSJ
The online self-tuning of PID parameters can be realized by International Conference on Intelligent Robots and Systems,
using the method of fuzzy reasoning, which not only possesses the Sendai, Japan, 2004, pp. 2973-78.
advantages of conventional PID control, but also prevails over the [3] T. Noritsugu, H. Yamamoto, D. Sasaki, and M. Takaiwa,
“Wearable power assist device for hand grasping using
latter in many performances, such as response time, flexibility,
pneumatic artificial rubber muscle,” in Proceedings of Annual
adaptability and control precision. But the tracking error of MCP is
Conference of SICE, Sapporo, Japan, 2004, pp.420-425.
bigger than that of CM and the tracking curves have jitters. The [4] E. J. Koeneman, R.S. Schultz, S. L. Wolf, D. E. Herring, and J.
main reasons are list as following: B. Koeneman, “A pneumatic muscle hand therapy device,” in
1. The friction force of MCP has bigger influence than that of Proceedings of the 26th Annual International Conference of
CM, because the main body of MCP is more complex than that of the IEEE EMBS, San Francisco, CA, USA, 2004,
CM. The influence can not be eliminated completely in the control pp.2711-2713.
system. [5] C. D. Takahashi, L. D. Yeghiaian, V. Le, R. R. Motiwala, and
2. In the process of rotation, angle sensor is not absolutely S. C. Cramer, “Robot-based hand motor therapy after stroke,”
accurate although filtering technique was employed. Brain, 2008, no.131, pp.425-437.
3. As an external disturbance, the user also affects the [6] N. Tsagarakis and D. G. Caldwell, “Improved modeling and
smoothness of robotic joints rotation. assessment of pneumatic muscle actuators,” in Proceedings of
the 2000 IEEE International Conference on Robotic and
5. DISCUSSION AND CONCLUSION Automation, San Francisco, CA, USA, 2000, pp.3641-3646.
[7] G. J. Silva, A. Datta, and S. P. Bhattacharyya, “New results on
the synthesis of PID controllers,” IEEE Trans. Autom. Control,
In this paper, a wearable rehabilitation robotic hand actuated by Vol. 47, no.2, 2002, pp. 241-252.
pneumatic artificial muscles is developed for the stroke patients. [8] K. H. Ang, G. Chong, and Y.Li, “PID control system analysis,
To realize the trajectory tracking control, PID and fuzzy PID design, and technology,” IEEE Trans. Control Sys. Technol.,
controllers were implemented and evaluated by experiments. The vol. 13, no. 4, 2005, pp.485-492
results show that the fuzzy PID controller has good performance. It [9] Yuzhi Wu, Haihua Wang, “Application of Fuzzy Self-tuning
is better than the PID controller in trajectory tracking. Besides, the PID Controller in Soccer Robot” IEEE International
designed device was capable of assisting the users to perform Symposium on Knowledge Acquisition and Modeling
grasping and releasing motions for repetitive hand therapy. The Workshop, 2008, 21-22 Dec. pp.14- 17
future work is the development of an intelligent adaptive controller [10] Kyoung Kwan Ahn, Dinh Quang Truong, Yoon Hong Soo,
to encourage the patient’s voluntary efforts during task-oriented “Self tuning fuzzy PID control for hydraulic load simulator,”
rehabilitation training. More experiments will be implemented to International Conference on Control, Automation and Systems,
improve the accuracy and stability of the angle/position control. 2007, 17-20 Oct. pp.345 - 349
Other control modes considering the measured force will be [11] Wei Ji, Qi Li, Bo Xu, “Design Study of Adaptive Fuzzy PID
developed for more flexible training ways. The current robotic Controller for LOS Stabilized System,” Sixth International
hand presented in the preliminary experiment can be further Conference on Intelligent Systems Design and Applications,
improved in mechanism and intelligence in the future. It is 2006. Volume 1, 16-18 Oct.pp.336 - 341
expected that this system will eventually: 1) empower patients to
choose when and where they want their exercise therapy; 2)
support therapists in a labor intensive task; 3) reduce dependency
on hospital resources; and 4) assist in re-integrating patients with

- 413 -

Authorized licensed
View publication stats use limited to: Huazhong University of Science and Technology. Downloaded on May 30,2010 at 14:11:06 UTC from IEEE Xplore. Restrictions apply.