An Adaptive Neural Network Control Method for Robotic Manipulators Trajectory Tracking

In engineering applications, the trajectory tracking control performance of a robotic manipulator is affected by many uncertain factors, such as disturbance and friction. It is difficult to establish an exact model for a robotic manipulator with unknown dynamics. This paper proposes an adaptive control algorithm based on RBF neural network for the manipulator trajectory tracking control. Firstly, a neural network is employed to off-line identify the unknown model of the manipulator system. Then, another RBF neural network is adopted to approximate and compensate the uncertain item in the nonlinear dynamic model respectively. Based on Lyapunov stability theory, the corresponding adaptive adjustment laws of neural network weights are derived, and a robust controller is designed to further reduce the approximation error of the model. At last, experimental results are performed to illustrate the trajectory tracking performance and anti-interference ability of the proposed control method.