Performance Estimation and Dimensioning of Team Size for Multirobot Patrol

The performance of multirobot patrolling teams heavily relies on the number of cooperative mobile robots that carry out the patrol mission. In the deployment phase, such systems are typically dimensioned to guarantee a given performance criterion. However, this is often done empirically using trial and error approaches. This article investigates the problem of estimating the performance of teams of robots in patrol missions, with the ultimate goal of providing the appropriate number of robots before the start of the mission. The authors show empirically that they can produce an upper bound of performance in four different environments, enabling them to define team size and guarantee that the time elapsed between consecutive visits to any location is less than a given value. The tradeoff in performance, team coordination, and gap between estimated and practical results in simulations and real-world experiments is analyzed.