A swarm control framework integrating embodied intelligence and multi-behavior coordination

To tackle the prevalent issues of limited coordination efficiency and delayed dynamic response in multi-agent systems operating under dynamic and resource-constrained conditions, this study proposes a cluster control framework that integrates embodied intelligence with behavioral diversity coordination. By introducing embodied intelligence theory, a real-time perception-decision loop is constructed between agents and the physical environment, thereby enhancing individual agents' adaptability to environmental disturbances and changing constraints. Combined with a behavioral diversity mechanism, a dual-policy fusion strategy is designed to exploit the complementary advantages of heterogeneous and homogeneous policy networks. A dynamic adjustment mechanism for the fusion coefficient is implemented to balance individual adaptability and group coordination. To evaluate the effectiveness of the proposed approach, a carrier-based ammunition transportation task is selected as a representative scenario. System performance is assessed under three conditions: no disturbance, deck swaying, and sudden obstacle interference. Experimental results indicate that the regulated introduction of behavioral diversity significantly reduces inter-agent collisions and improves overall coordination performance. This work contributes theoretical foundations and technical insights into multi-agent coordination in complex environments, while also supporting the intelligent restructuring of carrier-based ammunition logistics systems.