Path Planning for Multi-UAV in a Complex Environment Based on Reinforcement-Learning-Driven Continuous Ant Colony Optimization

Unmanned Aerial Vehicles (UAVs) are increasingly deployed in environmental monitoring, logistics, and precision agriculture. Efficient and reliable path planning is particularly critical for UAV systems operating in 3D continuous environments with multiple obstacles. However, single-UAV systems are often inadequate for such environments due to limited payload capacity, restricted mission coverage, and the inability to execute multiple tasks simultaneously. To overcome these limitations, multi-UAV collaborative systems have emerged as a promising solution, yet coordinating multiple UAVs in high-dimensional 3D continuous spaces with complex obstacles remains a significant challenge for path planning. To address these challenges, this paper proposes a reinforcement-learning-driven multi-strategy continuous ant colony optimization algorithm, QMSR-ACOR, which incorporates a Q-learning-based mechanism to dynamically select from eight strategy combinations, generated by pairing four constructor selection strategies with two walk strategies. Additionally, an elite waypoint repair mechanism is introduced to improve path feasibility and search efficiency. Experimental results demonstrate that QMSR-ACOR outperforms seven baseline algorithms, reducing average path cost by 10–60% and maintaining a success rate of at least 33% even in the most complex environments, whereas most baseline algorithms fail completely with a success rate of 0%. These results highlight the algorithm’s robustness, adaptability, and efficiency, making it a promising solution for complex multi-UAV path planning tasks in obstacle-rich 3D environments.