基于2D激光和TOF相机的室内环境3D建图算法

Indoor 3D Mapping Algorithm Based on 2D Laser and TOF Camera

室内环境3D建图技术在移动机器人领域得到了广泛的应用,但由于三维环境可能存在低纹理的情况,建图与定位的精度面临挑战。为此,提出了一种基于2D激光和TOF相机的室内环境3D建图算法,首先,使用Karto SLAM构建2D占用栅格地图;然后,移动机器人再次扫描三维环境,当移动机器人的行走距离差值、偏航角度差值或与上一次重定位时间间隔达到阈值,使用实时相关性扫描匹配算法进行重定位,同时TOF相机采集一帧三维点云数据;最后根据移动机器人系统坐标系的转换关系构建三维点云地图。由于点云地图的存储十分占用内存空间,当机器人位姿估计存在误差时,会导致点云地图明显重叠,将点云地图转换成了OctoMap。所提出的建图算法在不同的环境中进行了实验,并验证了该算法的有效性。

An indoor 3D mapping algorithm based on 2D laser and TOF camera is proposed in this paper.Firstly,Karto SLAM algorithm builds 2D occupancy grid map.Secondly,the operator controls the mobile robot to scan the 3D environment again.When the moving distance difference,the yaw angle difference or the time interval from the last relocalization of the mobile robot reaches the threshold,the mobile robot uses the real-time correlation scan matching algorithm is used for relocalization of mobile robot. At the same time,the TOF camera acquires a frame of 3D point cloud data.Thirdly,the 3D point cloud map is constructed according to the conversion relationship of the coordinate system.Finally,the point cloud map is converted into OctoMap because the storage of the point cloud map is very memory-intensive,and when there are errors in the robot’s poses,it will lead to obvious overlap of the point cloud map.