This paper addresses a largely ignored perspective in multiple robot observation of multiple targets, i.e. that of the evaders. We present a robust distributed approach to target motion control that utilizes cooperation to minimize the average observation time of the evaders. Targets actively communicate with each other to gain a much richer view of the environment than their immediate sensors provide, and use such information to generate better motion decisions from the group viewpoint. Targets relay information coming from other targets to ensure maximal data spread, and cooperate to escape the pursuers in certain favorable situations. Extensive testing has demonstrated the superiority of the approach over other simpler strategies. Furthermore, robustness with respect to key environmental conditions has been evaluated, and the results suggest the algorithm is a good candidate for real-world applications.
展开▼