Dynamic path planning method for headland turning of unmanned agricultural vehicles

Abstract

Automatic headland turning is an important part for an unmanned agricultural vehicle at farming work. Minimize the time and travel distance and maximize the operation efficiency of the vehicles during the turning are the concerns of the current focus. This research tries to present a dynamic headland turning path planning method based on an asymmetric switch-back turning method for four-wheeled vehicles. When an agricultural vehicle slips in the field, the algorithm will re-plan the route dynamically according to the real-time position of the vehicle to improve the operation efficiency and reduce the mechanical wear caused by frequently adjusting the steering wheel. The simulation results showed that when the row width was 2 m and some noises were added in the turning process, the trajectory length planned by the proposed algorithm was reduced by 31.42 % and the lateral deviation was reduced by 95.65 % compared with that planned by the traditional switch-back algorithm. In field experiments, the vehicle approached the next working path with an average lateral deviation of 2.0 cm at a speed of 0.5 m/s during a turn. The average length of the trajectory was 11.84 m, and the average time consumed was 28.4 s. Compared with the traditional switch-back turning, the average trajectory length, average lateral deviation and average cost time were shortened by 10.51 %, 78.75 % and 10.70 %, respectively. The proposed dynamic headland turning path planning method has a significant advantage over the traditional switch-back turning method in terms of both accuracy and time efficiency. The algorithm solves the problems that when the headland width is narrow, the vehicle is difficult to turn and deviates from the predetermined trajectory due to sideslips. Moreover, it ensures the shortest path and time for the agricultural vehicles to turn in the field, which greatly shortens the overall path length and time of its field operations.