关键词: 无人机（UAV）, 无人车（UGV）, 空地协同, 三阶段优化, 路径规划方法

Abstract:

Unmanned Aerial Vehicle-Unmanned Ground Vehicle （UAV-UGV） air-ground collaborative systems are widely applied and garner significant attention across various domains. UAVs offer advantages in high speed and maneuverability but suffer from limited endurance， while UGVs， though possessing relatively constrained mobility， provide substantial ground payload capacity and can serve as mobile relay stations for UAVs. Path planning for such collaborative systems presents a complex and challenging problem， aiming to generate feasible paths for both UAVs and UGVs to cooperatively accomplish missions. This paper comprehensively considers practical constraints frequently encountered in real-world scenarios， including speed， energy consumption， power limitations， and obstacles， overcoming the limitation of traditional air-ground systems which often focus on a single perspective （either airborne or ground-based）. We establish a task-oriented UAV-UGV air-ground collaborative system model and propose a three-stage-optimization collaborative path planning method. Specifically， we develop a Meta-learning Local-search Genetic Algorithm （MLGA）， combining meta-learning strategies with local search， to solve the first-stage global path generation problem within the collaborative model. For the second-stage UGV obstacle avoidance， we introduce a temporal particle A* algorithm， integrating an improved particle filter algorithm with an optimized temporal A* algorithm. Finally， addressing the third stage， we formulate a time-constrained UAV charging scheduling solution based on the distinct speed， energy， and power characteristics of UAVs and UGVs. The effectiveness of the proposed three-stage planning method is rigorously validated through simulations using virtual scenario data and real-world experiments.

Key words: Unmanned Aerial Vehicle (UAV), Unmanned Ground Vehicle (UGV), air-ground collaborative systems, three-stage optimization, path planning method