DRI-LOAM: A LOAM-Based Dynamic Objects Removal Strategy Using Range Images

The presence of dynamic points is unavoidable in scan data of urban environments, which creates unwanted traces in the map. These traces of dynamic objects may act as obstacles and thus impede good localization and navigation performances. This paper proposes a novel dynamic points removal strategy. Firstly, a LOAM based algorithm is used to generate a set of local maps. Then alignment is performed between query scan and corresponding local map to find suspected dynamic and static regions in the local map. Euclidean Clustering is further used to separate static and dynamic points in the suspected dynamic regions. Finally, a point cloud fusion algorithm is designed to integrate local maps and construct a clean global map. The algorithm shows good performance in low-speed moving scenes.