Deep Reinforcement Learning in Multi-UAV Air Combat Maneuver Decision-Making: A Review of Key Techniques in Practice and Future Prospects

The technology of deep reinforcement learning-based multi-UAV air combat maneuver decision-making is a hot research area in modern military studies. By combining the advantages of deep learning in handling high-dimensional complex data and the advantages of reinforcement learning in autonomous long-term planning, intelligent behaviors for air combat maneuver decision-making emerge. Aimed at providing practical optimization suggestions or basic entry-level guidance for researchers in this field, this paper focuses on the key technologies involved in multi-UAV air combat from a practical perspective, including improvements in deep reinforcement learning algorithms, the design of efficient training methods, and the construction of multi-UAV combat environments. The paper introduces and summarizes current mainstream methods and innovative technological achievements of them. Finally, it discusses future key research directions: centering on multi-agent collaborative combat for UAV swarms, focusing on the evaluation and construction of air combat scenarios in real battlefield environments, and developing comprehensive intelligent decision-making systems based on diverse decision-making methods. These developments are of significant importance for modern air combat advancements and achieving air combat superiority.