Density target assignment pursuit strategy for multi-agent pursuit-evasion games in bounded region

The pursuit-evasion problem has broad applications in fields such as intelligent transportation and drone formation, focussing on multi-agent coordination in complex environments. However, effectively coordinating multiple agents for target assignment and obstacle avoidance under dynamic uncertainty remains challenging. In this paper, we propose a distributed real-time algorithm based on the Artificial Potential Field (APF) method to address the pursuit-evasion problem with multiple evaders. The algorithm utilises a distributed agent system where each agent independently calculates and updates its position in real-time. To address the issue of multiple evaders within the perception range of pursuer, a density allocation method is introduced, combined with the Smoothed Particle Hydrodynamics (SPH) method. The SPH method can calculate the density of any point in the field based on the distance between pursuers, and then introduce the density into the calculation of evader's weights. In addition, the pursuit strategy in obstacles-prone environments is also investigated. The simulation results validate the effectiveness of the proposed algorithm, demonstrating that both the average capture time and success rate have been significantly improved. Furthermore, the proposed solution provides a robust framework for multi-agent pursuit-evasion games under dynamic conditions and uncertainty.