Based on the research of the traditional dynamics model and preview model, a kind of path tracking controller is designed in view of the theory of neural network and fuzzy control. Utilizing the self-learning and self-adjusting characteristics of neural network, combined with the advantages of fuzzy control, a kind of network is designed to predict the speed of vehicle and its output and lateral deviation are taken as the input of the fuzzy network to control the steering angle. Finally, the joint simulation by CarSim and Matlab/Simulink software is carried out and the actual vehicle experiment is done, having verified the effectiveness and accuracy of the designed controller. © 2019, Society of Automotive Engineers of China. All right reserved.