Abstract: In order to improve the seam tracking precision of the mobile welding robot, a control system was designed based on the RBF neural network and PLD control in this paper. Its structure was introduced and the corresponding dynamics model was also established. The RBF neural network structure and the design method of controller were emphatically discussed. In order to improve the control performance, the gradient information was on-line identified by neural network as well as the proportion, integral and differential coefficients can be adjusted according to the gradient information. The simulation results show that the control method can realize complex trajectory tracking. In the seam tracking process, mobile welding robots run smoothly with high tracking precision.