基于多帧局部地图的室内TSDF建图算法

Indoor TSDF Mapping Algorithm Based on Multi-frame Local Maps

ICP是移动机器人室内建图的常用算法,通过不断地匹配相邻帧来完成机器人定位与地图构建过程。PL-ICP是ICP的优化版本,比起普通的ICP能实现更快的收敛速度和更高的精度。但不管是ICP或是PL-ICP,由于新一帧的位置只与上一帧有关,导致误差会不断累积,从而降低最后得到的地图的质量。论文提出一种局部地图匹配的方法,即通过累积多帧历史帧得到当前局部地图,并让新一帧与局部地图进行PL-ICP匹配的方法。实验证明这样做能够有效地减小匹配过程中的误差累积,从而提高机器人定位与建图的精度。在建图方面,论文采用TSDF建图方法,TSDF相比于普通的占用栅格建图方法能够得到更加准确的地图。

ICP is a common algorithm for indoor mapping of mobile robots.It completes the process of robot positioning and map building by continuously matching adjacent frames.PL-ICP is an optimized version of ICP,which can achieve faster convergence speed and higher accuracy than ordinary ICP.But whether it is ICP or PL-ICP,because the position of the new frame is only related to the previous frame,errors will continue to accumulate,thereby reducing the quality of the final map.This paper proposes a method of local map matching,that is,the current local map is obtained by accumulating multiple historical frames,and the new frame is matched with the local map by PL-ICP.Experiments have proved that this can effectively reduce the accumulation of errors in the matching process,thereby improving the accuracy of robot positioning and mapping.In terms of mapping,this article adopts the TSDF mapping method,which can get a more accurate map than the ordinary occupancy grid mapping method.