Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Proceedings of the 7th JFPS International Symposium on Fluid Power, TOYAMA 2008 September 15-18, 2008 P1-47 DEVELOPMENT OF GAIT TRAINING SYSTEM USING BI-ARTICULAR MUSCLE MODEL Yoshiyuki SHIBATA*, Motoki TAKAGI*, Tasuku MIYOSHI*, Ryota AOYAMA* Shin-ichiroh YAMAMOTO* and Yukio KAWAKAMI* * Graduate School of Engineering, Functional Control Systems Shibaura Institute of Technology 307 Fukasaku, Minuma-ku, Saitama-City, Saitama, 337-8570 Japan (E-mail: [email protected]) ABSTRACT The purpose of this study was to develop a pneumatic gait training system which was consisted by Mckibben type actuators to achieve the normal gait patterns of lower limb joint angular displacements. These actuators were arranged as human musculoskeletal system; mono- and bi-articular muscles. The system provided a proportional directional control valve which controlled antagonistic actuators, and this system had PID controller and joint angular displacement feedback circuit. In the results, the output of the knee joint movement pattern had well validation for the input signal which was recorded from natural treadmill walking in healthy subjects. This system made the orthosis reproduce like the human natural walking. In conclusion, we structured a fundamental bi-articular muscle type of the gait training system using Mckibben type actuators. KEY WORDS Bi-articular muscle, Mckibben Actuator, Gait training system INTRODUCTION The value of partial body weight support treadmill training (BWSTT) for disabled individuals has been well established, especially when initiated soon after an injury. Wernig et al. [1] had demonstrated that 25 out of 33 incomplete spinal cord injured persons learned to walk independently at the end of 3-20 weeks (median 10.5) with partial BWSTT. However, this treatment presents special challenges for the therapist. The passive moving of a disabled person's legs is ergonomically difficult since the individual being treated cannot support and/or move his legs by himself. In order to improve this therapy, Colombo et al. [2] developed a driven gait orthosis (DGO) that can be used on patients with varying degrees of paresis or spasticity for up to one-half hour. 505 Dietz et al. [3] using this gait training system for paraplegic patients, and they suggested the afferent input from hip joints is important for the leg muscle activation during locomotion. In the previous studies, we had constructed the two types of gait training systems using the Mckibben type of pneumatic actuators to achieve the repetitive physiological gait patterns [4-5]; one was mono-articular muscle model [4], the other was bi-articular muscle model [5]. The Mckibben type actuator had similar mechanism to human muscle contraction for a force generating, and it has high degree of freedom of design and contributes to orthosis lightening. In addition, both of our system has some advantages over the partial BWSTT on land; the system is electrically and mechanically safe since it consists of pneumatic Copyright © 2008 by JFPS, ISBN 4-931070-07-X actuators to prevent short circuits and a burn, and the main frame is an exoskeleton type of the hip-knee-ankle orthosis which is used in general gait training. According to Miyoshi et al. [4], they had developed a robotic gait trainer in water (RGTW) and had demonstrated its efficacy which could reduce the enhanced hip extensor muscle activities, since the water walking needs propulsive forces against the water resistance [6]. In this respect, the bi-articular muscle model could reduce the hip extensor muscle activities [5], since the bi-articular muscles had unique function in force-directing because of their ability to simultaneously regulate or tune adjacent joint torques [7]. The bi-articular muscles are through the two joint, and are arranged in a pair of antagonists; e.g. biceps brachii and triceps brachii, rectus femoris and biceps femoris. The synergistic activities of one antagonistic pair of the bi-articular muscles and two antagonistic pair of the mono-articular muscles might be able to reproduce by single command signal informing force direction via a simple spinal level neural network proposed [8]. To our knowledge, however, there is no gait training system using bi-articular muscle model, because the bi-articular muscle model is redundant and torque shearing problem for two joints is infinite combination. The purpose of this study was to develop the bi-articular muscle model type of the pneumatic gait training system for achieving the repetitive physiological normal gait patterns of lower limb joint angular displacements, and to design the specific controller for bi-articular muscle model based on PID. SYSTEM DESIGN Pneumatic Control Figure 1 represents the gait training system. The orthosis drives by contraction of the Mckibben type actuators, and we applied antagonistic arrangement of the Mckibben type actuators like human muscles. So, we used proportional directional control valve(servo valve). This valve has 1 input port and 2 output ports which attached to the Mckibben type actuators posterior and anterior. Therefore it’s very suitable for control by antagonistic actuators. Mckibben type actuator Figure 2 shows the Mckibben type actuator. This actuator generated force by contraction. The outer-tube of mesh texture transformed a radial force into an axial force when inner rubber-tube generated force by expansion. Contraction ratio was 30% on natural length. 750 mm Joined Natural length Shortening Figure 2 The mechanism of Mckibben type actuator Figure 1 Gait training system schematic diagram Copyright © 2008 by JFPS, ISBN 4-931070-07-X Joined 506 parameter sets P:1.0, I:0, D:0, to all actuators. The hip joint angle output value was twice as value than the target value, and there were phase shifting(figure 5). The mono-articular muscle model had time-delay, and couldn't reach the maximum target angle. Walking program and control interface We used a Digital Signal Processor(DSP) for input and output of signals. This equipment has A/D and D/A converter, and control program run the Simulink of Matlab(The MathWorks). Figure 3 shows the block diagram of servo system. Input data was joint angle of gait cycle by healthy people. Hip and knee joint angle of orthosis got feedback which detected by the potentiometer. The PID controller value decided voluntarily from the experiment. The DSP provides interface function which called Virtual Console. It function is very useful which is able to create one's own GUI, and load parameter from Simulink of Matlab. The user is able to change the parameters without Simulink of Matlab. C (s) kI s T in (s) + - kp + + + G (s ) T out (s ) kD s Figure 4 Left leg of orthosis 㻢㻜 㼐㼑㼓㼞㼑㼑㻌㻌㻌㼒㼘㼑㼤㼠㼕㼛㼚 Figure 3 Block diagram of servo system. T in (s ) : Target angle, T out (s ) : Controlled angle, C (s ) : PID controller, G (s ) : Servo valve control voltage gain. 㼔㼕㼜㻌㼛㼡㼠㼜㼡㼠 㼔㼕㼜㻌㼕㼚㼜㼡㼠 㻠㻡 㻟㻜 㻝㻡 㻜 The following show equipment list. a) Mckibben type actuator: Hitachi Medical Corp. b) Servo valve: MPYE-5-1/8-HF-010-B, FESTO Corp. c) DSP: ADX5430, A&D Co. Ltd. Orthosis overview Figure 4 shows the orthosis overview. The left leg of orthosis fixed to the frame, and suspended. At first, we build up this system only unilateral leg of orthosis on a trial basis. The Mckibben type actuator attached to the anterior and posterior of each joint. The orthosis drove because these actuators contract alternately. 㻙㻝㻡 㻙㻟㻜 㻙㻠㻡 㻙㻢㻜 㼐㼑㼓㼞㼑㼑㻌㻌㼒㼘㼑㼤㼠㼕㼛㼚 㻢㻜 㼗㼚㼑㼑㻌㼛㼡㼠㼜㼡㼠 㼗㼚㼑㼑㻌㼕㼚㼜㼡㼠 㻡㻜 㻠㻜 㻟㻜 㻞㻜 㻝㻜 RESULTS 㻜 It is possible for proportional directional control valve to control the both agonistic/antagonistic Mckibben type actuators. Thus, it can be decreased the number of control valves to the number of actuators. We achieved simplification of the pneumatic pipe arrangement in this system. This experiment was performed the same parameter between bi-articular muscle model and hip mono-articular muscle model. The PID controller 㻙㻝㻜 㻜 㻞 㻠 㻢 㻤 㻝㻜 㻝㻞 㻝㻠 㼟㼑㼏 Figure 5 Joint angle variation during orthosis walking. Knee joint angle data(bottom), and hip joint angle data(upper). Target values were hip and knee joint angle for healthy people(one gait cycle:0.5Hz, solid line: target joint angle, break line: controlled angle value). 507 Copyright © 2008 by JFPS, ISBN 4-931070-07-X 6. T. Miyoshi, T. Shirota, S. Yamamoto, K. Nakazawa and M. Akai, Effect of the walking speed to the lower limb jointangular displacements, joint moments and ground reaction forces during walking in water, Disability and Rehabilitation, 2004;VOL.26, NO.12, 724–732. 7. G. J. Van Ingen Schenau, From rotation to translation: Constraints on multi-joint movements and the unique action of bi-articular muscles, Human Movement Science, Volume 8, Issue 4, August 1989, Pages 301-337. 8. M. Kumamoto, T. Oshima, and T. Fujikawa, Bi-articular muscle as a principle keyword for Biomimetric motor link system, Microtechnologies in Medicine & Biology 2nd Annual International IEEE-EMB Special Topic Conference on, 346-351, 2002. DISCUSSION The hip joint angle response was larger than target angle value, it was not appropriate parameter of servo valve control voltage gain. The reason of phase shifts might be overlarge proportional value of PID controller. The bi-articular muscle model was through the two joint, but the angle feedback was performed only hip joint angle. It wasn’t optimal follow-up control. The human walking had coinstantaneous rotation knee and hip joint. Hence, we have to determine the PID controller value and feedback parameter for each muscle models. CONCLUSION We developed a fundamental gait training system with position feedback and PID controller functions, drove by pneumatic Mckibben type actuator. This system might be possible which reproduce the movement like the human natural walking. FUTURE WORK In this study, it was difficult for the bi-articular muscle model to control by hip joint angle feedback. We will determine about the optimal PID controller value of mono-articular muscle model and the appropriate control parameter for the bi-articular muscle model. We have to superstruct control system which mono and bi-articular muscle models will be possible to produce their coordination. REFERENCES 1. Wernig A, Muller S, Nanassy A, Cagol E. Laufband therapy based on 'rules of spinal locomotion' is effective in spinal cord injured persons. Eur J Neurosci 1995;7:823-9. 2. G. Colombo, M. Joerg, R. Schreier, and V. Dietz, Treadmill training of paraplegic patients using a robotic orthosis, Journal of Rehabilitation Research and Development Vol. 37 No. 6, pp.693-700, 2000. 3. V. Dietz, R. MuÈller and G. Colombo, Locomotor activity in spinal man: significance of afferent input from joint and load receptors, Brain, 125, 2626-2634, 2002. 4. T. Miyoshi, K. Hiramatsu, S. Yamamoto, K. Nakazawa and M. Akai, Robotic gait trainer in water: Development of an underwater gait-training orthosis, Disability and Rehabilitation, 2008; 00(0): 1–7. 5. S. Yamamoto, T. Miyoshi, T. Komeda, K. Hiramatsu, K. Nakazawa, and M. Akai, Development of pneumatic gait assist system, Complex Medical Engineering, 2007. CME 2007. IEEE/ICME International Conference on, 1337 – 1340. Copyright © 2008 by JFPS, ISBN 4-931070-07-X 508