Abstract
This study aims to design and experimentally implement a nonlinear fault tolerant control scheme for shape memory alloy actuator employed for a robotic manipulator. In the proposed scheme, a super twisting sliding mode controller as a baseline controller is firstly developed based on a reduced order model of the overall system. This controller robustly controls the manipulator position in fault-free conditions of the shape memory alloy actuator. Simultaneously, a nonlinear observer is introduced to estimate perturbations and actuator faults. If the estimated perturbation exceeds a predefined threshold for unmodeled system dynamics, a potential actuator fault is detected. When actuator fault is detected, the decision mechanism activates and utilizes observer data to reconfigure the controller, thereby compensating for the fault. The stability of the controller has been demonstrated in the presence of input constraints using the Lyapunov method. Experimental results for healthy actuator demonstrate that the baseline controller maintains robustness against uncertainties and unmodeled dynamics without requiring perturbation estimates. Consequently, the perturbation information enables both fault detection and compensation within a simplified control architecture. Comparative results with a passive fault-tolerant controller indicate the superior performance of the suggested scheme in compensating for faults, uncertainties, and unmodeled dynamics.