A DISCRETE-TIME NEURAL CONTROL METHOD FOR MANIPULATOR TRACKING PROBLEMS

This paper introduces a neural-network-based control method for robot trajectory tracking problems. The proposed approach uses an inverse dy namics neural network model to generate nominal control signals to dri ve the robot to follow the desired trajectory. In addition, a second n eural network is employed to estimate the inaccuracy associated with t he inverse dynamics neural network model. Based on such estimations, a feedforward compensation strategy is then adopted to reduce the influ ences of neural model inaccuracy Computer simulations are presented to demonstrate the effectiveness of the proposed approach.