Autonomous Defense of Unmanned Aerial Vehicles Against Missile Attacks Using a GRU-Based PPO Algorithm

This paper introduces a method enabling unmanned aerial vehicles (UAVs) to autonomously defend themselves against incoming missiles during military missions. A simulation environment for UAV defense against incoming missiles was created by constructing motion models for a UAV, a missile, and infrared decoys in three-dimensional space. A UAV defense strategy generation algorithm based on gated recurrent unit (GRU) and proximal policy optimization (PPO) is proposed, which effectively addresses the problem of low survival rates when under attack from enemy missiles. Specifically, the algorithm estimates the current true state based on current and historical observations and makes decisions based on the state estimate, effectively solving the non-Markov problem caused by the lack of infrared decoy information in the observations. The experimental results indicate that this method provides an effective defense strategy, combining evasion maneuvers with infrared decoys to effectively evade incoming missiles.