Indoor Multi-Robot Cooperative Mapping Based on Geometric Features

Abstract

To improve the efficiency and accuracy of the indoor multi-robot cooperative mapping, the extraction method of overlapping regions between each local map based on the geometric features is proposed. First, the geometric features of local maps are extracted and fitted to represent the semantic information of the indoor building structure. Second, the attribute similarity evaluation, the topology similarity evaluation, and the ICP evaluation are used to extract the most similar frame pair between each local map. Third, the local map is merged by overlapping area registration to construct a global map. The proposed method is verified in three different real-world scenarios, the experimental result shows that the efficiency and accuracy of the cooperative mapping method are improved compared with the single-robot mapping method and the existing cooperative mapping method.