Trajectory planning for UAV swarm based on reinforcement rearning: a review

Reinforcement learning (RL) refines policies through continuous interaction with the environment, offering a model-free paradigm for unmanned aerial vehicle (UAV) trajectory generation. A systematic review devoted to RL-based cooperative path planning of multi-UAV systems is, however, still missing. To close this gap, we present a comprehensive survey of the relevant literature. First, we formalize the problem statement and identify the key challenges inherent in large-scale swarm trajectory optimization. Next, we summarize the fundamentals of RL and taxonomize its representative algorithms with emphasis on their specific adaptations to UAV path planning. Finally, we synthesize the achieved advances and remaining limitations, and outline prospective research directions. The delivered insights are intended to serve as a reference for future theoretical investigations and practical implementations.