Abstract: The energy consumption of a UAV(Unmanned Aerial Vehicle) determines the length of its operational cycle. To address the issue of low communication-to-energy consumption ratio, a reinforcement learning-based UAV path planning solution is proposed to reduce energy consumption while maintaining high communication quality. The continuous flight space is divided into multi-layer two-dimensional grids to facilitate the generation of UAV state points, and a reward function based on communication quality parameters and energy consumption parameters is established. The Q-Learning algorithm is employed to learn and obtain the path with the optimal communication-to-energy consumption ratio. Experimental results show that the path planned by this learning model can achieve a higher communication-to-energy consumption ratio, demonstrating its practical value.

Key words: route planning, Q-Learning algorithm, unmanned aerial vehicle