移动机器人同时定位与建图室内导航地图研究

针对当前视觉同时定位与建图(simultaneous localization and mapping,SLAM)生成的点云地图不能满足路径规划和导航的需要,提出一种室内移动机器人的导航地图制备方法.首先,通过SLAM估计相机位姿,后端优化后生成室内场景的三维点云地图;其次,根据地面移动机器人的运动约束及结构特点分情况讨论,推导点云相对于地面的二维坐标,同时对点云进行地面与障碍的分离、截取与筛选;最后,根据栅格占据状况有序构建出导航地图.实验结果表明,基于点云坐标的障碍物截取准确度高于地面拟合截取方法,所建地图精度与完整度均高于传统方法.室内移动机器人能基于该地图进行路径规划与导航.

Research on Simultaneous Localization and Mapping of Indoor Navigation Map for Mobile Robot

Aiming at the current point cloud map generated by Simultaneous Localization And Mapping (SLAM) that cannot meet the needs of path planning and navigation, a method for preparing a navigation map for an indoor mobile robot is proposed. Firstly, the camera pose was estimated based on SLAM, and a 3D point map of the indoor scene was created after back-end optimization .Secondly, according to the motion constraints and structural characteristics of the mobile robot on the ground, the two dimensional coordinates of the point cloud relative to the ground were deduced. Then separated the ground and obstacle, intercepted and selected the point. Finally, a navigation map was constructed in an orderly manner according to the grid occupation status. Experimental results show that Obstacle interception accuracy based on point cloud coordinates is higher than that of ground fitting interception method, the accuracy and completeness of the built map are higher than traditional methods. The indoor mobile robot can perform path planning and navigate based on the map.