Ant Colony Optimization applied to the Multi-Agent Patrolling Problem

Patrolling an environment involves a team of agents whose goal usually consists of continuously visiting its most relevant areas as frequently as possible. For such a task, agents have to coordinate their actions in order to achieve optimal performance. Current research that tackles this complex multi-agent problem usually de nes the environment as a graph, so that a wide range of applications can be dealt with, from computer network management to computer games and vehicle routing. In this paper, we consider only instances of the multi-agent patrolling problem where all agents are placed on the same starting node. These instances are often encountered in robotics applications, where e.g. drones start to patrol from the same area. The Ant Colony Optimization is adopted as the solution approach to these problem instances. Two novel ACO algorithms are proposed here, in which several ants' colonies are engaged in a competition for nding out the best multi-agent patrolling strategy. Experimental results show that, for three graph topologies out of the six which were evaluated, one of our ACO techniques, GU/AA, signi cantly outperforms the reinforcement learning technique proposed by Santana et al. (2004) (i.e. GBLA), irrespective of the number of the involved patrolling agents. For the other graph topologies, GU/AA approaches the results obtained