【摘要】 为了提高覆盖路径规划任务的性能，提出了一种基于深度强化学习的多尺度地图无人机覆盖路径规划方法。首先对地图进行中心化和不同尺寸映射的处理，其次加入了Luong注意力机制，最后设计不同权重的奖励函数。实验表明改进后的无人机覆盖路径规划方法可以提高无人机对目标区域的覆盖范围以及成功着陆率。更多还原

【Abstract】 To improve the performance of overlay path planning tasks, a multi-scale map UAV coverage path planning method based on deep reinforcement learning is proposed. Firstly, the map is centralized and mapped with different sizes. Secondly, the Luong attention mechanism is added to extract features of more interest on the map. Finally, the reward function with different weights is designed. The experiments show that the improved UAV coverage path planning method can improve the coverage and successful landing rate of the UAV to the target area.更多还原