A Bio-Inspired Adaptive Formation Architecture Based on Multi-Agents with Application to UAV Swarm

This paper introduces a novel architecture for bio-inspired adaptive formation and system stability in unmanned aerial vehicle (UAV) swarms during complex operations. Utilizing a Multi-Agent System (MAS) framework, this architecture incorporates various bio-inspired mechanisms tailored to specific tasks, enabling UAVs to observe and respond independently to the actions of neighboring agents. This decentralized approach promotes effective cooperation without centralized control, employing evolutionary rules for dynamic adaptation. The architecture comprises a three-tier framework that manages capability transfer and decomposition, integrating both bottom-up and top-down methodologies. This structure aligns operational requirements with logical architecture rules and supports the definition of state machines for UAV agents. Inspired by biological swarms such as birds, ants, and wolves, the UAV control rules are designed based on the agent structure of the UAVs to guide swarm behavior in different scenarios. Simulations in a virtual environment demonstrate the swarm's capability to autonomously execute ‘search and firefighting’ missions. The results, analyzed using flight mission planning software, confirm the feasibility of the adaptive formation strategy and validate the generated flight trajectories against mission constraints. The primary innovations of this study include the development of a decentralized decision-making process for UAVs, simulating real unmanned systems' decision modes, such as path planning and cooperative tactics. Furthermore, the use of multi-agent software in constructing virtual mission scenarios facilitates concept verification and supports enhanced simulations with actual aerial vehicles, bridging the gap between theoretical models and practical applications.