No abstract

This paper studies a dynamic swarm versus swarm unmanned aerial vehicle (UAV) combat problem and proposes a self-organized offense–defense confrontation decision-making (ODCDM) algorithm. This ODCDM algorithm adopts the distributed architecture to account for real-time implementation, where each UAV is treated as an agent and able to solve its local decision problem through the information exchange with neighbors. At each decision making step, the swarm seeks an optimal target allocation scheme and each UAV further selects the corresponding behavioral rules, leading to emergent offensive and defensive behaviors. Therefore, the offense–defense confrontation decision-making process is divided into the target allocation decision based on distributed consensus-based auction algorithm (CBAA) and social-force-based swarm motion decision. An offense–defense preference is introduced to the target allocation optimization model, providing the tactics options for UAV to adopt more offensive or more defensive posture. On the basis of classic collective behaviors of cohesion, separation and alignment, a combat stimulus is considered to drive UAV towards the assigned target. Finally, simulation experiments are carried out to verify the effectiveness of the ODCDM algorithm, and analyze the influences of the external deployment and internal tactics on the combat results.