Design and control of automatic finger extensor based on iris mechanism for hand rehabilitation system

Abstract

The impairment of motor function in stroke patients cause them to be paralysed. With the extensive rehabilitation training and exercise, the probability for stroke patient to regain movement is high. At early stage, most of the stroke patient cannot extend their finger because of the weakness in the muscle. At this rate, their fingers are always in flexed condition. The therapist has to extend their fingers to prevent the muscle hardened. However, limitation of the therapist has become a crucial problem and with the help from the robotic rehabilitation, the rehabilitation become easier and helpful. Finger extensor can help the patients to perform the exercise precisely and repeatedly. For finger extensor, the mechanism needs to provide variable diameter opening to extend the fingers and the opening need to be controlled, so that it follows the desired trajectory. This research focus on the development of an automatic finger extensor based on iris mechanism and its controller based on Sliding Mode Control-Function Approximation Technique (SMC-FAT) based adaptive control. Motion simulation studies and Finite Element Analysis (FEA) has been conducted on the proposed automatic finger extensor. The prototype of the iris mechanism has been fabricated and the test shows that it has worked successfully as required. The formulation of a Sliding Mode Control-Function Approximation Technique (SMC-FAT) based adaptive controller for proposed automatic finger extensor based on iris mechanism has been presented. In this research, the controller is able to cater friction uncertainty and external force from the patients. In this research, friction uncertainty is solved using FAT expression where FAT expression issued to represent the uncertainties. In FAT methods, Radial Basis Function Neural Network (RBFNN) is used as the basis function. The stability of the controller can be proven using Lyapunov function. Simulation test and hardware experimental test using MATLAB, Simulink and Real Time Window Target have been conducted to verify the effectiveness of the controller. In the simulation, the results show that the controller successfully compensate the uncertainties and external force with average Root Mean Square of 1.35 mm.