In order to realize the automatic navigation of agricultural vehicles, ultra wide band (UWB) wireless positioning technology was used to obtain the current position of the vehicle in real time, and the path tracking research was carried out. In view of the inconsistency between the theoretical steering motion model and the actual steering motion model caused by factors such as the vehicle's own processing, assembly errors and tire sideslip, in order to improve the control accuracy, the vehicle's steering motion model was identified based on the least squares method, and the turning motion models at different linear speeds were obtained. At the same time, in order to solve the problem that some of the control variables calculated by the conventional pure tracking algorithm cannot make the vehicle steering system respond, resulting in the decline of the path tracking accuracy, a two-stage pure tracking algorithm considering lateral deviation and lateral deflection angle was proposed. And the strategy of hysteresis switching of setting transition lag was proposed for how to switch between the two stages. The “S” type path tracking test in the simulation environment showed that when the driving speed was 0.6m/s, the average lateral deviation of the pure tracking algorithm with a fixed look-ahead distance was 0.5238m,and the two-stage pure tracking algorithm was 0.3616m, and its tracking accuracy was improved by 30.9%, which had better path tracking performance than that of the pure tracking algorithm with fixed look-ahead distance. The hysteresis switching stratege was adopted to reduce the two stages mutation rate from 2.18% to 1.16%, and the suppression effect was improved by 46.8%.