A fault-tolerant modular control approach to multi-robot perimeter patrol

The use of large scale multi-robot systems is motivated by a number of desirable features, such as scalability, fault tolerance, robustness and lower cost with respect to more complex and specialized agents. This work is focused on a a behavior-based approach to the problem of the multi-robot border patrolling, in the framework of the Null-Space-based Behavioral control (NSB); it is based on two previous works of the same authors, where the feasibility of the approach is demonstrated. Namely, a few aspects of the approach, not yet tackled in previous works, are investigated: its robustness to faults of individual agents, its capability of managing large numbers of robots, the possibility of adding new tasks in the framework of the multi-robot patrolling problem. Along these directions, our approach has been validated in simulation with a large number of robots and sudden faults as well as experimentally on a team composed by three Pioneers 2-DX robots.