抄録
In security robotics, it is a critical problem to intercept an intruder before it arrives at its target place. To protect valuable objects from the intruder, a security zone is observed and patrolled cooperatively by multiple robots. Until now, previous studies related to patrol have been concentrated on planning paths to satisfy some criterion such as visit frequency and there was no approach to analyze danger of unguarded areas. For this reason, we suggest an algorithm for discovering vulnerable areas when an intruder appears. Vulnerable areas are a set of nodes that none of robots can intercept the intruder if it appears on that area. We partitioned a security zone to n (the number of robots) regions and assigned them to each robot to patrol with cyclic paths. Finally, changes of vulnerable areas according to the changes of locations of robots along time were showed. The proposed algorithm can be used as a tool to evaluate patrolling methods.