An enhanced computed-torque control scheme for robot manipulators with a neuro-compensator

In this paper, a novel control scheme is proposed to enhance the conventional computed-torque control structure of robot manipulators. The control scheme is based on the combination of a classical computed-torque control as a feedforward structure and a neural network as a compensation structure. The resulting control scheme has a simple structure with improved robustness. Since the neuro-compensator has good adaptability, accurate knowledge of both the robot dynamic parameters and structure are not required in advance, and large parameter variations during operation can also be compensated for. This special control algorithm also has the implementation advantage in that the neuro-compensator is independent of the feedforward control loop and, therefore, a multi-rate sampling structure for the whole control system can be applied. Simulation studies verify the effectiveness of the proposed control method.