农用轮式移动机器人相对位姿的求解方法

Pose Sensing of Agricultural Wheeled Mobile Robot

在基于单目视觉的自主导航中,由于农用轮式移动机器人相对于跟踪路径位姿的传统求解算法,往往存在忽视图像中各像素点权重不同和计算效率不理想等缺陷,因此,针对农田环境特点,在分析地面上直线路径透视成像特性的基础上,提出了一种农用轮式移动机器人相对位姿的求解方法。该方法首先建立起被跟踪路径在图像平面上的像素坐标与机器人相对位姿间的关系方程,然后结合Hough变换的思想直接求出位姿值。实验结果表明,该方法不仅可以有效地弥补传统算法的不足,而且测量精度也与当前其他类似研究的水平大致相当。

Generally, one camera is used in the autonomous navigation of agricultural wheeled mobile robot. Conventional algorithm of pose sensing, however, ignores the fact that pixels in digit image possess the different weight, and its compute time can not be optimized easily. In order to overcome these shortcomings, the perspective characteristic of the linear guiding road on the ground was analyzed carefully, then the relationship between the road coordinate in the digit image and the robot pose was established, so the relative pose could be obtained directly with the Hough transformation. Based on the prototype of agricultural wheeled mobile robot, the experimental results showed that the precision derived from this method is similar to those demonstrated by the other same type of researches.