: Patrolling tasks in multi-robot systems are essential for applications such as surveillance and monitoring, where minimizing the time between visits to any given location is critical. This paper investigates fault-tolerant redistribution strategies for multi-agent patrolling systems in partitioned environments using Unmanned Au-tonomous Vehicles (UAVs). We propose Heuristic Meeting-based Patrolling (HMP), a novel distributed and fault-tolerant patrolling algorithm. Building on the Heuristic Conscientious Reactive (HCR) strategy and incorporating periodic synchronization meetings, HMP enables decentralized coordination and dynamic fault recovery through minimal communication. UAVs exchange information at shared meeting points, enabling detection of failures and redistribution of responsibilities. We evaluate HMP and its simplified variants in various simulated environments using the Multi-Agent Exploration and Patrolling Simulator (MAEPS). The results demonstrate that HMP offers strong performance under both normal and fault conditions, comparative to state-of-the-art patrolling strategies in terms of idleness metrics. However, we also identify limitations in meeting scheduling under certain fault conditions, which can cause cascading failures. Based on these findings, we discuss potential improvements for future work, including enhanced meeting scheduling and adaptive partitioning strategies.

Abstract as published.