To improve the navigation path tracking controlling performance of small crawler tractors which are widely used in traditional orchards, a navigation path tracking control method based on the virtual radar model was developed. Drew on the human experience of driving, referred to the principle of radar scanning and the idea of image recognition, the virtual radar model was constructed to generate the virtual radar map. The virtual radar map was used to describe the position relationship between the vehicle and the path, and the corresponding driving operation instructions were generated by the deep neural network classification. The typical U-shaped path of orchard operation was taken as an example to carry out simulation and real vehicle test, and the simulation results showed that the proposed method can accurately achieve the navigation path tracking control; orchard test results revealed that when the vehicle speed was 0.36m/s and 0.75m/s, maximum lateral error of path tracking was 0.150m and 0.191m, the average lateral error was 0.031m and 0.051m, the standard error was 0.025m and 0.036m, respectively; compared with the designed fuzzy control method, the maximum lateral error was reduced by 15.73% and 36.33%, the average lateral error was reduced by 27.91% and 19.05%, and the standard error was reduced by 21.88% and 28.00%, respectively. The results demonstrated that the navigation path tracking control method based on the virtual radar model presented high path tracking accuracy and driving stability, which met the actual operational needs of orchards. It provided a way of thinking to solve the navigation path tracking control problem.