Abstract: It firstly built the path planning environment model of agricultural vehicle from the working space model and the constraints of updating and maneuvering performance of the vehicle. It used the machine learning algorithm based on the improved artificial potential energy to realize the optimization of the path planning of agricultural vehicle combining with the path planning environment model of agricultural vehicle and the obstacle detection module. Matlab simulation results show that the path planning system of agricultural vehicles can identify the location and boundary information of obstacles,and can quickly develop the optimal flight path to guide agricultural vehicles to reach the target point smoothly.