Deep Reinforcement Learning-Based UAV Path Planning Algorithm

In the unmanned aerial vehicle (UAV)-enabled wireless charging sensor networks systems, the UAV is able to replace the base station for data collection. Each deployed sensor and the data it collects have their own unique values. Therefore, the UAV should collect as much data as possible from different sensors. We formulated the UAV trajectory planning problem with the goal of maximizing the amount of data collected. Due to the time-varying character of the WSNs and the immediacy of UAV path planning, we apply deep reinforcement Learning (DRL) technique to tackle this challenge. In order to collect as much data as possible, we used a Double Deep Q-Network (DDQN) with a data collection action selection strategy. The simulation results show that compared to the DQN algorithm, the DDQN algorithm guided drone access instructions can collect more data.