Grid Map Merging with Insufficient Overlapping Areas for Efficient Multi-Robot Systems with Unknown Initial Correspondences

초록

This paper addresses a grid map merging problem in multi-robot systems with unknown initial correspondences. If robot-to-robot measurements are not available, the only way to merge the maps is to find and match the overlapping area between maps. But, if the overlapping area is insufficient, the performance of the existing map merging methods degenerates. This paper proposes a new map merging algorithm using the Radon transform, which can be successfully conducted with relatively insufficient overlapping areas. Because the Radon transform can extract abundant geometric information of a map according to rotation and translation, the map transformation matrix can be accurately computed by matching the sinograms producted by the Radon transform. Experiments with a public dataset and a real multi-robot system showed that our algorithm using sinograms can accurately merge the maps, and the required overlapping area is smaller than other map merging methods with similar computation time.