The Open System Setting in Timed Multiagent Patrolling

Several techniques have been proposed in the last few years to address the multiagent patrolling task. They share the assumption of a closed system setting (the set of agents present in the system is constant, no agent joins or leaves), which is a strong requirement and limits the applicability of multiagent patrolling models. In this article, we propose to revisit some of the techniques proposed in the literature to adapt them to the open society setting, and to compare their performances on a simple scenario where an agent decides to quit the patrolling task.