Cooperation through mediation in multi-robot pursuit games

This paper presents a control strategy to the pursuit problem of an intruder with evasive as well as nonevasive behavior. In present approach, the decision-making process is sub-divided into individual agents' ranking over the available task space and group level inference. A novel ranking mechanism embedded within individual members of pursuit team enables robots to determine their own rankings on available task space, taking into account agent's internal as well as external states. While such decision engine provides autonomy to individuals at every decision cycle, the central mediation unit prevents any contingent mission failure that may incur due to uneven distribution of agents over available task space, via mediating the individuals' decisions. Properties that can be attributed to one such system are investigated. Simulation results and performance analysis of the proposed approach for evasive as well as nonevasive intruder are demonstrated.