Research on Multi-Aircraft Trajectory Planning Based on Reinforcement Learning

This study explores the application of deep reinforcement learning algorithms in the field of air traffic, particularly focusing on intelligent trajectory planning for multiple aircraft. By using the K-means clustering algorithm to identify trajectory patterns in the Shanghai terminal area and employing a grid method for three-dimensional airspace modeling, the trajectory planning space for aircraft was significantly reduced. Finally, the DDQN algorithm was used to achieve trajectory planning for aircraft within the Shanghai terminal area. The research covers terminal area airspace modeling, trajectory pattern recognition, and aircraft trajectory planning. By configuring a simulation environment based on the Shanghai terminal area airspace model, the effectiveness of the DDQN algorithm in improving the efficiency and safety of autonomous aircraft scheduling was validated, demonstrating the great potential of deep reinforcement learning algorithms in solving multi-aircraft trajectory planning problems.