Aligning methods for visual landmark-based maps

When having a multi-robot system in which each robot contructs its own local map, it can be necessary to perform the fusion of these local maps into a global one. The Map Fusion problem involves the consecution of two different tasks: Map Align- ment and Map Merging. The Map Alignment consists in computing the transformation, if existent, between the local maps. In this way, all the observations will be referenced to a common global frame. In the Map Merging stage, a global map is constructed from the local maps by integrating their information. This pa- per is focussed on the first step: Map Alignment. Par- ticularly, a collection of aligning algorithms is eval- uated in order to select the method that obtains the best results in terms of accuracy and stability. The experiments are performed in a multi-robot system, in which each robot constructs its own local map in- dependently. These maps are visual landmark-based and the mapping algorithm used is FastSLAM.