A New Entrapment Based Invader Capture Strategy for Multi-robot Surveillance Systems

Abstract

This study presents a new entrapment-based invader capture strategy for multi-robot surveillance systems. Most approaches have been focused on enclosing an invader using a particular formation. However, in terms of physical robot capability, it is hard to capture the invader practically. In this study, the invader has its escape cut off while multiple robots stand against it at some critical points on the invader’s path. The strategy consists of four parts. A graph is constructed on a given map for the Voronoi roadmap using the Voronoi diagram. Since constructed graph has a lot of nodes and edges, the thinning process is performed while reducing the computational load. Subsequently, candidate nodes are extracted on the whole map. If the invader escapes along the shortest path to the exit, the critical cluster is selected from the candidate nodes by checking the reachability of arrival times to the points. The proposed strategy is applied to a SLAM map using multiple robots. The results show that critical points are properly found, and all robots capture the invader successfully.