Intelligent Pursuit and Evasion Decision-Making in Active Defense Scenarios

In the complex environment of modern aerial warfare, an attacking missile attempts to successfully penetrate defenses and strike a target aircraft, while a defensive missile intercepts the attacking missile to protect the target. Meanwhile, the target aircraft itself possesses intelligent evasion capabilities, employing smart maneuvering strategies to evade the attacking missile's pursuit. In this triadic confrontation within an active defense scenario, intelligent strategy decisions of all participating agents are critical to the outcome of the engagement. In the dynamically evolving battlefield, improving the penetration capability of the attacking missile, enhancing the coordination of multi-agent decision-making, and strengthening the evasion capability of the target aircraft are key challenges. To address these challenges, the objective of this study is to enhance the attacking missile's penetration ability in an active defense scenario, particularly in the presence of interference from a defensive missile, while also improving the coordination and collaborative effectiveness of multiple agents. Therefore, this work proposes an improved Multi-Agent Proximal Policy Optimization (MAPPO) algorithm that incorporates a Transformer Encoder. By introducing the Transformer Encoder, the algorithm's representation capacity and decision-making precision are significantly enhanced, addressing the limitations of traditional MAPPO in processing complex environmental information, penetrating defensive missiles, and generating intelligent evasion strategies for the target aircraft.