This paper proposes a new algorithm for manipulator system-an adaptive sliding mode robust control algorithm based on RBF neural network. Based on the traditional sliding mode control method, the RBF neural network is used to approximate the manipulator model information and external interference. We established the system model of the six-degree-of-freedom manipulator XIOPM developed by our research group. In order to verify the effectiveness and superiority of the algorithm in the simplest possible case, we took the models of the first two joints and performed it through MTALAB. The simulation results are consistent with our expectations. Compared with the movement of the manipulator under traditional sliding mode control, our method can not only make the actual output trajectory of the manipulator system converge to the desired trajectory at a relatively faster speed, but also reduce chattering to a large extent. The control algorithm reduce the disadvantages of traditional sliding mode control. Its good tracking performance and tracking accuracy make this manipulator system well controlled.