Multi-robot map alignment in visual SLAM
This paper focusses on the study of the Map Alignment problem in a multirobot SLAM context. Given a team of robots, we consider the situation in which each robot is building its own local map independently. These maps are landmark-based and three-dimensional. The local maps built by the different robots will have different reference systems. At some point, it may be interesting to express all maps in the same reference system. In that way, the maps can be fused into a unique global one. This is known as the map fusion problem and can be tackled by dividing the problem into two subproblems: map alignment and map merging. In this paper, we concentrate on the map alignment problem which consist on computing the transformation between the local maps so that we have one reference system. We therefore evaluate a set of aligning methods. These methods establish correspondences between each pair of maps and compute an initial estimate of the alignment. Finally, we apply the least squares minimization to obtain a more accurate solution. We also concentrate on the case in which we want to align more than two local maps. In this case, the alignment must be globally consistent. For the experiments, real data are used. Each robot extracts distinctive 3D points from the environment with a stereo camera. Then, the robots use these observations as landmarks to build their maps while simultaneously localize themselves. The SLAM problem is solved using the FastSLAM algorithm. The movements of the robots occur only in 2D plane, so although the maps are 3D landmark-based, the alignment takes place only in 2D.