Research on Multi-UAV Path Planning Based on the MATD3-APF Algorithm

Multi-UAV systems are widely used in disaster response, environmental monitoring, and reconnaissance. However, safe and efficient 3D path planning remains challenging due to static and dynamic obstacles and the need to ensure inter-UAV safety. Traditional Artificial Potential Field (APF) methods are prone to local minima and strong parameter dependence even though computationally efficient and physically interpretable. This paper proposes an enhanced MATD3-APF framework that integrates APF with the Multi-Agent Twin Delayed Deep Deterministic Policy Gradient (MATD3) algorithm under a centralized training and decentralized execution (CTDE) scheme. In this framework, APF provides instantaneous geometric guidance for obstacle avoidance, while MATD3 adaptively optimizes the repulsive coefficients associated with obstacles. Furthermore, a Geodesic Spherical Normalization (GSN) mechanism is introduced, which applies bounded rotations on the directional manifold to preserve vector norms. Simulation results demonstrate that the proposed approach enables multi-UAV systems to generate shorter trajectories and achieve faster arrival times.