FINNCH: Cooperative Pursuit Navigation for a Pursuer Team to Capture a Single Evader in Urban Environments

Abstract

The development of a cooperative pursuit strategy for capturing escaping criminals or dangerous animals in urban public safety emergencies is becoming increasingly in demand. An ideal strategy should consider both the encirclement needed to prevent criminals from evading and the distance that pursuers need to move. This article proposes a fine-grained navigation network-based cooperative hunting (FINNCH) method. A fine-grained navigation network is created to provide detailed information about the traversability in urban areas. Three interaction rules inspired by biological behaviors in nature are introduced to achieve dynamic cooperation between pursuers. An heuristic search strategy is used to guide the pursuers toward potentially good directions, which consequently reduces the search effort. Two spatial constraints, namely, the direction centrality constraint (DCC) and encirclement distance constraint (EDC), are then constructed to evenly distribute the pursuers around the evader. Several experiments are conducted to evaluate the effectiveness and efficiency of the proposed method. The results show that FINNCH can provide navigation for multiple pursuers in complex urban environments comprised of roads, meadows, trees, water, and buildings. These findings point to the promising future of FINNCH for practical applications.