Distributed Multi-UAV 3D Trajectory Planning Based on Multi-Agent Deep Reinforcement Learning

With the widespread application of unmanned aerial vehicles (UAVs) in smart city scenarios under the Internet of Everything, multi-UAV collaborative trajectory planning faces multiple challenges including cost constraints, collision risks, and complex action spaces. To address these issues, a distributed multi-UAV 3D trajectory planning collaborative optimization method is proposed, which enhances system efficiency through joint optimization of UAVs energy consumption and Age of Information (AoI). Firstly, an improved genetic algorithm-based optimization mechanism is designed to determine the number of UAVs by introducing threat factors and composite reward factors. Then, a multi-agent deep reinforcement learning (MADRL)-enabled collaborative planning framework is established, modeling the trajectory planning problem as a multi-objective Markov decision process (MDP) to achieve autonomous trajectory optimization in complex urban environments through distributed policy networks. Finally, the system implements adaptive speed adjustment for UAVs using a greedy algorithm to improve resource utilization. Simulation results demonstrate that compared with baseline algorithms, the proposed method achieves superior performance in reducing energy consumption, enhancing AoI, and optimizing trajectory planning strategies, validating its effectiveness in intelligent urban air mobility systems.