Saturated adaptive back-stepping control for robot manipulators with RBF neural network approximation

Abstract

Aiming at solving the actuator saturation problem for the robot manipulators, a saturated adaptive back-stepping controller with neural network approximation is proposed. Different from the traditional back-stepping controllers, a class of saturation function and projection-type adaptation are applied to make the torque control inputs bounded. In the meantime, a Radial Basis Function (RBF) neural network based approximator is designed to replace some complex expressions in the control law, which facilitates the practical implementation of the proposed controller. In addition, explicit and strict stability analysis is given via Lyapunov's direct method, which shows that all the signals of tracking error of the system are uniformly ultimately bounded. Finally, simulation comparisons indicate that the proposed controller results in a more satisfactory tracking performance, and can suppress the initial sharp oscillation of the torque control inputs for each joint effectively.