An Evolutionary Reinforcement Learning Approach for Autonomous Maneuver Decision in One-to-One Short-Range Air Combat

This paper presents an evolutionary reinforcement learning approach based on Deep Q Networks to address the maneuver decision challenge of unmanned aerial vehicles (UAV) in short-range aerial combat. The proposed approach aims to improve the UAVs’ autonomous maneuver decision process and generate a robust policy against alternative enemy strategies. The training process involves parallel training of multiple workers, evaluation of models at regular intervals, selection of the best model, testing the best model against enemy policies, and updating the pool of enemy strategies. The proposed method continuously improves the trained models and generates more robust policies with higher win rates than standard reinforcement learning techniques or k-level learning approaches.