针对多关节机械手的鲁棒跟随控制器设计问题，提出了一种新的机械手神经网络自适应滑动模控制器设计方法，机械手的动力学非线性假设是完全未知的。在提出的控制结构中，高斯径向基函数神经网络用于在线补偿机械手的动力学非线性，参数学习律由稳定性理论得到。给出了系统稳定性和参数收敛性的证明。最后提出方法的可行性通过仿真得到验证 Aiming at the design problem of robust follow controller of multi-joint manipulator, a new design method of manipulator neural network adaptive sliding mode controller is proposed. The assumption of dynamic nonlinearity of manipulator is completely unknown. In the proposed control structure, Gaussian radial basis function neural network is used to online compensate the dynamics nonlinearity of the manipulator. The parametric learning law is obtained from the stability theory. The proof of system stability and parameter convergence is given. The feasibility of the proposed method is verified by simulation