基于各向异性空间投影的地铁列车底部巡检机器人避障研究

Research on Obstacle Avoidance of Robot Arm for Subway Train Bottom Inspection Based on Anisotropic Spatial Projection

针对地铁列车底部的非示教场景,提出一种基于各向异性的空间投影的点云分割新算法。算法针对使用立体相机获取的三维点云数据,将点云数据投影到机器人坐标系下的不同平面上,然后利用聚类算法获得目标的分割区域,再映射回点云空间计算出相应的包围盒。在整列地铁车厢底部进行的多次实验验证取得了预期的效果,证明了算法的有效性,可以实现地铁列车底部复杂环境下的高效规划避障。

A new point cloud segmentation algorithm based on anisotropic spatial projection for non teaching scenes at the bottom of the subway train is proposed.This algorithm aims at the 3D point cloud data obtained by stereo camera,projects the point cloud data to different planes under the robot coordinate system,and then the segmentation area of the target is obtained by using the clustering algorithm,and then the corresponding bounding box is calculated by mapping back to the point cloud space.Many experiments have proved the effectiveness of the algorithm,which can achieve efficient obstacle avoidance under the complex environment of the bottom of the subway train.