Research on UAV Path Planning Based on Particle Swarm Optimization and Soft Actor-Critic

This study proposes a groundbreaking path planning method in UAV path planning in complex 3D environment, I-PSO-SAC, which seamlessly integrates Particle Swarm Optimization (PSO) with the Soft Actor-Critic (SAC) reinforcement learning algorithm to address these intricate challenges. The innovative approach leverages PSO-SAC to initially generate feasible paths, followed by advanced interpolation sampling techniques to refine and optimize the path quality. Experimental results show that I-PSO-SAC improves the convergence speed by 42.57% compared to the traditional SAC algorithm in static scenarios and 51.01% in dynamic scenarios. In addition, the method significantly improves the smoothness of paths, outperforming other deep reinforcement learning algorithms.