Entropy-Based Key Node Identification for Group-Splitting Swarm Control of Vicsek Model

The critical challenge of group-splitting in self-organizing particle swarms operating in confined environments is addressed in this paper. We propose an entropy-guided control framework built upon an enhanced Vicsek model, which incorporates anisotropic sensing, limited field-of-view, and leader influences. By constructing a dynamic interaction network, our method leverages a novel structural entropy metric to identify topologically critical nodes within the swarm. Sparse environmental guidance, selectively applied to these key nodes, then propagates through the swarm's intrinsic local interactions. This enables coherent obstacle avoidance and robust re-cohesion with minimal intervention. Simulations demonstrate our framework's superior obstacle avoidance and control efficiency compared to baseline strategies, showcasing its effectiveness in maintaining swarm integrity.