Indoor mapping using low-cost MLS point clouds and architectural skeleton constraints

MLS (Mobile Laser Scanner) offers adequate mobility and the ability to obtain accurate environmental data. However, it still faces some challenges, such as the heavy reliance on high-resolution LiDAR and the poor reliability of the mapping process, which limit its effectiveness in engineering projects. Therefore, we propose a novel, automatic and efficient indoor mapping method that utilizes low-cost MLS point clouds and architectural skeleton constraints. First, we determine the initial localization using designed architectural skeleton feature patterns and descriptors. Second, we register adjacent scans with different registration rules due to different abundance of architectural skeletons. Third, we correct cumulative error for both loop and non-loop areas. Last, we stitch scans into the indoor point cloud map, and build the novel lightweight point cloud map. Experiments are carried out in three typical large-scale scenes inside buildings, and results show that our method can produce accurate indoor maps automatically and efficiently.