A Multi-robot Exploration Approach Based on Distributed Graph Coloring

The assignment of target points to multiple robots in exploration missions is a combinatorial problem that is known to be NP-Hard. Negotiation-based methods, such as auctions, have been extensively investigated in the literature, but the provided solution is usually sub-optimal. In this study, we propose an approach which models the multi-robot exploration task as a distributed graph coloring problem, where the targets to be visited in the environment are the nodes of a graph that must be colored and each robot represents a different color to be used in the coloring process. As result, the coloring approach is able to minimize the exploration costs of the team, since it captures the synergy among targets and assigns them to a small number of robots. The proposed algorithm was tested in several experiments performed in environments with cluttered and clustered arrangements of targets, and compared the outcomes with those of a distributed auction approach on the same setting. The results show that our method compares favorably, and indicate that distributed coloring is a promising approach for reducing the costs for exploration task using multi-robot systems.