CADR-RAS Webinar: Distributed Multi-Agent Coordination for Energy-Constrained Patrolling

In this research, the application we consider relates to the use of sensor-equipped unmanned air and ground vehicles to support the protection of ground-based assets on the battlefield. The purpose of sensing is to detect potential threats operating in the vicinity of these assets. We frame the problem as a multi-agent patrolling problem where the objective of the autonomous agents is to frequently and repeatedly visit key locations within a defined area in order to minimise the time any location remains unvisited (i.e., its idleness). This capability is critical for continuous surveillance and timely threat detection in dynamic environments. We assume the agents operate with limited energy capacity, requiring periodic returns to a base for recharging, which directly impacts patrol duration and coverage. The project focuses on adapting and extending existing online, distributed approaches to operate effectively under such constraints by developing heuristic decision-making algorithms that account for energy availability and idleness. A number of these approaches involve coordination strategies that seek to improve team performance without introducing inefficiencies. We evaluate algorithm performance across varied graph topologies in a simulation framework, and our analysis provides insights regarding the impact of algorithm characteristics on performance. The ultimate aim of the research is to compare context-aware patrolling algorithms for energy-constrained environments, with future work extending these methods to also address communication constraints for improved operational resilience.

We look forward to having you attend the event!