Stochastic Multi-Agent Patrolling Using Social Potential Fields
Evgeny A. Shvets · 2015
In this paper we consider a task of decentralized, multi-agent patrolling of a continous outdoor terrain. We propose an algorithm that efficiently operates under the condition of low communication throughput and is robust to the failure of one or more patrolling agents. The solution is based on Social Potential Fields and is easily extensible to allow other types of behavior. We describe an agent-based simulation system and use the obtained results to show how the patrolling algorithm should be altered to be effective on different types of terrain. Several techniques to increase the efficiency of patrolling are provided.