No abstract

This paper proposes a neural network-based continuous finite-time tracking controller for the robust high-precision control of robotic systems under model uncertainty, external disturbance, and actuator saturation. First, a fast nonsingular integral terminal sliding mode (FNITSM) surface is adopted to ensure singularity avoidance and fast finite-time convergence. Considering the presence of model uncertainties and external disturbance, the fully adaptive radial basis function neural network (ARBFNN) is used to approximate and compensate for the unknown dynamic model. Then, a novel continuous fast fractional-order power (CFFOP) approach law is explored to increase the convergence rate and eliminate chattering in the FNITSM control. Meanwhile, the approach law relaxes the requirement on the exact information of the upper bound of the disturbances and their time derivatives. Besides, an actuator saturation compensator (ASO) is proposed to compensate for the limited control input. The stability and finite-time convergence of the proposed controller are analyzed using the Lyapunov theory. Finally, comparative simulations of both the numerical and application examples are conducted to verify the effectiveness of the proposed control schemes, indicating that the CFFOP approach law and ASO can be used effectively for robotic systems.

No abstract