Adaptive RBFNN Slide Mode Control of a Robotic Manipulator System with Actuator Failure

In the paper, an adaptive radial basis function neural network (RBFNN) slide mode control of a robotic manipulator system with actuator failure is proposed, which facilitates rapid adjustment and tracking of the target value or reference trajectory by introducing a sliding surface. This sliding surface demonstrates effective responsiveness to external changes in comparison to the traditional PID control. The incorporation of the powerful function approximation capability inherent in the RBFNN renders it well-suited for practical applications characterized by parameter uncertainty or challenges in constructing an accurate model. By combining RBFNN with sliding mode control (SMC), the system surpasses the necessity for an accurate system model prevalent in traditional SMC. The stability based on Lyapunov theory is analyzed. MATLAB is employed to conduct simulations, and the comparison with conventional PID algorithm verifies the effectiveness of the proposed method. It illustrates that the proposed method excels in terms of tracking accuracy, robustness, and rapid convergence considering actuator failure.