Study of the Vertical Pursuit Point Design for TTP-Based Air Combat Autonomous Decision-Making

The tactical pursuit point (TPP) method is an innovative strategy in the current unmanned aerial vehicle (UAV) close-range air combat. However, the existing TPP method heavily relies on expert knowledge (EK) in the generation of its vertical pursuit point. To address this issue, two vertical pursuit point design methods are proposed, one based on a combination of EK and reinforcement learning (RL) and the other based solely on RL. And the twin delayed deep deterministic policy gradient algorithm (TD3) is utilized to form the RL parts of the proposed methods. The simulation results show that compared with the traditional EK-based method, both the proposed methods can enrich the vertical maneuvers of the aircraft and improve the win rate of air combat. In particular, the RL-based method shortens the time to win, increases the win rate, and effectively improves the autonomy in the decision-making process of air combat.