基于双目视觉的主从式果园作业车辆自主跟随系统设计

将黑白棋盘格特征板固定于引导车辆作为跟随特征,通过固定于跟随车辆的双目视觉系统,获取黑白棋盘特征板上各角点的三维信息,分析该信息,得到引导车辆相对于跟随车辆的车辆间距、横向偏移及航向偏角,利用该导航信息实现跟随车辆的自动跟随,建立了主从式果园作业车辆的跟随系统。双目视觉系统在不同光强下静态性能测试结果表明,双目系统生成导航参数偏差与日间光强无明显相关性,体现出双目系统对外界光环境变化有良好的适应性。在果园环境下,对该系统进行性能测试,结果系统导航信息生成频率为4.0 Hz,横向偏移平均偏差为3.68 cm,车辆间距平均偏差为8.24 cm,航向偏角平均偏差为2.32°,表明所构建的跟随系统能实现对引导车辆的自动实时跟随。

Design of autonomous following system for master-slave vehicles operating in orchard based on binocular stereo vision

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.