A Local Search Approach for Improving Multi-Robot Routing in Exploration Missions

Cooperative multi-robot exploration is one of the fundamental problems in mobile robotics. A typical exploration problem is that in which robots have a model of the environment and they must visit a set of virtual, fixed target points in order to perform some task there. The assignment of multiple robots to target points in order to minimize the team costs, such as traveled distances, is an NP-Hard problem. Negotiation-based methods such as auctions have been widely used in literature, because they are easy to implement and present low communication and computational requirements. However, due to their simplicity, these methods produce sub optimal solutions. This paper presents a local search algorithm which can be combined with auctions to improve the solution quality of the assignment without increasing the communication requirements. The approach proposed in this paper is compared with three other auction-based allocation methods proposed in literature and the experiments showed that the local search was able to outperform the other approaches, minimizing the team costs in terms of traveled distances during the mission.