Multi-robot exploration framework using robot vision and laser range data

We propose in the following paper to tackle the multi-robot exploration problem. Our approach does not presuppose the availability of maps of the environment. We first describe a laser based on-line map building associated with a landmark based methodology to fuse team robots observations in the general case where no relative poses are given. We then explain how to assign the next best target in case of multi-robot exploration. Details are also given on our 3D landmark recognition system and on how to deal with the multiple landmark images search during exploration.