Abstract:

To address the mission planning problem of aerial recovery in complex environments， a mission planning model incorporating recovery time windows has been established， and a hierarchical framework is adopted to mitigate the coupling characteristics between task assignment and trajectory planning. First， a trajectory pre-planning method based on Pruned Probabilistic Roadmap （P-PRM） and Dubins curves is designed to analyze UAV recoverable intervals and integrate them into the mission planning process. Subsequently， a Distributed Particle Swarm Optimization （DPSO） algorithm is implemented for recovery task assignment， taking into account the characteristics of aerial recovery mission. Furthermore， to satisfy both spatiotemporal and obstacle avoidance constraints， a trajectory library is constructed utilizing four distinct homotopy-based trajectory deformation methods. The trajectory feasibility is analyzed in Configuration-Time （CT） space， and optimal homotopy parameters are efficiently determined through binary search optimization. Experimental validation demonstrates the effectiveness of the proposed integrated planning and replanning methodology. In comparison with centralized Particle Swarm Optimization and Genetic Algorithm approaches， DPSO demonstrates superior performance in search efficiency and convergence rate. Relative to conventional trajectory planning methods in large-scale scenarios， P-PRM exhibits enhanced planning speed while maintaining compact trajectory lengths. The implemented homotopy deformation method effectively satisfies both obstacle avoidance and expected flight time constraints within the trajectory planning framework.

Key words: aerial recovery, task assignment, Particle Swarm Optimization (PSO), Probabilistic Roadmap (PRM), homotopy trajectory