It is of important significance to diagnose study the fault for motion system of the omni-directional soccer robot to enhance its robustness. On basis of analyzing the fault types of motion system, the method based on probability neural network (PNN) for Omni-directional soccer robot was proposed. Among the method, the Kalman filtering method was used to process the residuals signal generated by the robot’s velocity in real time, and then the fault of the motion system was diagnosed accurately through the PNN. A simulation example is given to demonstrate the validity of the method. 