A vision-based guidance system for an agricultural vehicle

This paper introduced the vision-based guidance system of agricultural vehicle. According to the method extracting the guidance parameters by Hough transform, a new method based the center line detecting and erosion algorithm for the path recognition was designed. The guidance control system of agricultural vehicle was analyzed, a PD controller was developed to minimize the path error. The vehicle´s guidance accuracy was tested on paths constructed of common blue artifical plant, the guidance system guided the tractor automatically through the straight path, path tracking performance was observed. The largest error of the lateral displacement is 7.5cm, the average error of the lateral displacement is 1.54cm and the largest error of the heading displacement is 1.443deg, the largest error of the heading displacement is 0.05deg, when the vehicle was tested on the straight path at a speed of 5km/h. The image processing average time was 168ms per image. Those results can meat the requirement of practical application.