Collaborative decision-making for UAV swarm confrontation based on reinforcement learning

Abstract

With the advancement of unmanned aerial vehicle (UAV) technology, research on adversarial interactions within UAV swarms has gained significant attention domestically and internationally. However, the existing decision-making algorithms are primarily tailored to homogeneous UAV swarm adversarial scenarios, facing challenges such as complex reward function design and limited decision-making timeliness when applied to more intricate scenarios. This article investigates the real-time control decision-making issues in UAV swarm adversarial interactions. First, an adversarial simulation environment for UAV swarms is constructed, which effectively unifies the environment and state representation, enhancing the response speed of our UAVs. Second, a distributed UAV swarm collaborative control algorithm based on multi-agent reinforcement learning is proposed, and an effective sparse reward function is designed to guide UAVs in adversarial gaming, making the UAV strategies more aggressive, enhancing the adversarial intensity, and further optimizing the control strategy to meet real-world demands better. Finally, the real-time performance and scalability of the proposed method are validated through simulations.