Aiming at the characteristics of nonlinear,strong coupling and time-varying in the field of redundant manipulator kinematics,a method of nonlinear fitting and system identification based on radial basic function(RBF) neural network is adopted.RBF neural network,with the capabilities of good approximation,fast convergence,strong nonlinear processing ability and etc.,can effectively resolve the redundant manipulator inverse kinematics problem.Meanwhile,the center parameters of RBF network are selected by genetic algorithm to improve the performance and efficiency.The 6-DOF manipulator kinematic modeling and simulation is implemented with Matlab and the calculated results verify the effectiveness of the method proposed in this paper.