Abstract:A checkerboard plate was fixed on the back part of the master-vehicle as a guidance feature for the slave-vehicle. The three-dimension(3D) information of each corner point on the checkerboard plate can be easily obtained by the stereo vision system which was fixed on the front of the slave-vehicle. Through the analysis of the 3D information of each corner point, the distance, the relative lateral offset and heading angle can be obtained between the two vehicles. These information can be used to guide the slave-vehicle to follow the master-vehicle autonomously. According to this method, a master-slave following system was built for vehicles operating in orchard. The test results of stereo vision system performance in different light environment showed that the navigation parameter deviation is not associated with the light intensity during the day, so the stereo vision system have good adaptability to the natural light environment. The performance of the master-slave vehicles system was tested in an orchard. The update rate of the navigation information was 4.0 Hz. During the operation, the average deviation of lateral offset, the distance, heading angle was 3.68 cm, 8.24 cm and 2.32° respectively. The experimental results showed that the slave vehicle can follow the master-vehicle in real time successfully.