The accuracy of the kinematics model of agricultural machinery affects the accuracy and stability of navigation control. In order to improve the accuracy of the path tracking controller of agricultural machinery, a design method of navigation controller based on the motion characteristics of agricultural machinery was proposed. The method was mainly an improvement of the traditional two-wheeled vehicle kinematics modeling method. Aiming at the shortcomings of the small-angle approximation replacement (the direction angle was equal to yaw angle) of the traditional two-wheeled vehicle model, the method of adding the slip angle was used to optimize the kinematics modeling process of agricultural machinery. The same control method (state feedback control) and different kinematics models were used to design controllers for control experiments. When tracking a straight path, the slip angle had little effect on the model accuracy, and the introduction of slip angle can affect the tracking accuracy of agricultural machinery to a certain extent. When tracking a curved path, the slip angle had a great impact on the change of the direction angle, which can greatly affect the path tracking accuracy. Using a tractor equipped with automatic navigation equipment as an experimental platform for field experiments. The experimental results showed that the maximum lateral error of straight driving was 0.0454m, the absolute average error was 0.0149m and the standard deviation was 0.0119m. The maximum lateral error of curve driving was 0.1613m, the absolute average error was 0.0688m and the standard deviation was 0.0434m. The data showed that the path tracking controller designed based on the proposed kinematics model optimization method can improve the linear path tracking to a certain extent, and can greatly improve the curve tracking accuracy.