Finding optimal routes for multi-robot patrolling in generic graphs

Abstract

Multi-robot patrolling is a problem that has im- portant applications in security and surveillance. However, the optimal task assignment is known to be NP-hard. We consider evenly spacing the robots in a cyclic Traveling Sales- man Problem (TSP) tour or partitioning the graph of the environment. The trade-off in performance, overall team travel cost and coordination is analyzed in this paper. We provide both a theoretical analysis and simulation results across multiple environments. The results demonstrate that generally cyclic- based strategies are superior, especially when small teams are used but at the expense of greater team cost, whereas partitioning strategies are especially suitable for larger teams and unbalanced graph topologies. The reported results show that graph topology and team size are fundamental to determine the best choice for a patrol strategy. I. INTRODUCTION Advances on autonomous mobile robots have been evident in the last couple of decades. In particular, the patrolling problem with a team of cooperative agents has received much focus. The problem, which is also known as repet- itive sweeping, has unquestionable utility in society and finds its applications in surveillance systems, infrastructure security and inspection, search and rescue, mine clearing, military operations, environmental monitoring, intelligent transportation, household cleaning, and several other areas. Being monotonous, these tasks may also be dangerous. Thus, improving safety and reducing fatigue is a major advantage of multi-robot patrolling systems. In this context, robots are required to continuously travel in the environment, and the key challenge is to design efficient routes in order to optimize a certain performance criterion. Like most existing work in the literature (1-4), it is assumed that robots are homogeneous, travel with constant speed, and are expected to visit every strategic position of the environment. Therefore, having adequate sensing range, complete coverage of the environment is achieved by visiting all the important locations in the area. Despite several recent contributions to the problem, one question still remains open: what is the optimal patrol strategy for a given generic environment using R robots? This work was supported by a PhD scholarship (SFRH/BD/64426/2009), the CHOPIN research project (PTDC/EEA-CRO/119000/2010) and by the Institute of Systems and Robotics (project PEst-C/EEI/UI0048/2011), all of them funded by the Portuguese science agency Fundac ¸ ˜ ao para a Cie a