Performance-based adaptive modulation of resistance in hand rehabilitation system using finger-extensor mechanism

Abstract

In robot-aided therapy, control algorithms provide minimal assistance to patients during therapy to encourage active participation from them. This is accomplished by triggering assistance based on patient participation. However, such control strategies fail to account for changes in patient performance within a single exercise session. This leads to a slacking response from the patient which impedes recovery. Furthermore, resistive therapy, which helps patients in regaining motor function more effectively has not been widely implemented in the area of hand rehabilitation. The objective of this research is to develop a performance-based impedance control algorithm for modulation of resistive force exerted on the patient during a single therapy session. First, the system model of the 1-DOF finger extensor rehabilitation machine was developed. Then a performance-based impedance control law was designed with the capability of using force exerted by the patient during therapy as a basis for modulation of resistive force applied by the machine on to the patient. Stiffness parameter of the controller and the consequent force applied on the patient was modulated with each change in patient performance, as measured by a Force Sensing Resistor (FSR). A Graphical User Interface was also developed to provide real-time feedback of patient performance. The system model developed in the first step was validated through simulation and hardware experimentation. Implementation of the control algorithm was carried out on the real system and resistive therapy experiments were performed with a healthy subject. Performance of the developed controller was evaluated by drawing a comparison between the reference force generated by the algorithm and actual force output on the subject during three resistive therapy sessions. Root Mean Squared Error of 0.9875 was obtained which shows that the developed mathematical model represents the real behaviour of the physical system closely. Mean absolute error over the three resistive therapy sessions between the reference force and the force exerted by the machine on the subject was found to be 0.843 N whereas the relative error was 3.72%. Based on the experimental results, it is proven that the developed control strategy is able to change the control parameter within a single therapy session and modulate resistive force exerted on the subject throughout the session. Hence, the controller is successful in avoiding slacking behaviour during therapy.