Abstract:A self-adaptive control method based on RBF neural network for manipulator was put forward in order to solve parameters time-variation, strong coupling and high nonlinear of manipulator system model as uncertain factors. This method was made use of self-adaptability, fault-tolerant, parallel processing and nonlinear mapping ability of RBF neural network, so as to realize the control without accurate manipulator model information. According to the simulation experiment under Matlab/Simulink environment, it shows that this method can realize the position tracking control of SCARA manipulator. By timely correction of network parameters through the control algorithm, the arbitrary trajectory tracking control of the nonlinear system is realized with good control quality.