Frequency-Based Multi-Robot Fence Patrolling

Patrolling—sometimes referred to as repeated coverage—is a task where points in a target work-area are repeatedly visited by robots. The patrolling algo rithm can be optimized towards different performance criteria, such as freque ncy of point visits, minimization of travel time to arbitrary points of interest, etc. Most existing work provides patrolling solutions for closed-polygon work-areas, in which a topologically-circular path is used to guarantee optimal performance. I n contrast, we focus in this paper on patrolling along an open polygon, e.g., a tw oended fence. Here, there are inherent challenges to maintain uniformity of point visit frequencies, and other performance criteria. We introduce two co ordinated multi-robot fence-patrol techniques: The synchronized andsynchronized-overlap methods. We show that in general, the synchronized approaches to multirobo patrolling outperform the individual, unsynchronized methods. We then a nalyze the performance of the synchronized methods in depth, with respect to d ifferent performance goals, and investigate key trade-offs.