A research team led by Associate Professor Sun Guodong from Beijing Forestry University's School of Information Science and Technology (School of Artificial Intelligence) has published a paper titled "Scheduling Drone and Mobile Charger via Hybrid-Action Deep Reinforcement Learning" in a top journal IEEE Transactions on Mobile Computing, a CCF-A ranked journal.

There has been growing interest in using chargers to extend the operational longevity of UAVs (drones). In this paper, the research team explores a charger-assisted drone application where a drone observes points of interest while a mobile charger moves to recharge its battery. The study focuses on optimizing the routes and charging schedules of both the drone and mobile charger to maximize observation utility in the shortest possible time while ensuring continuous drone operation. In this system, the drone and mobile charger cooperate to complete tasks, but their discrete-continuous hybrid actions present significant computational challenges. To address this issue, the researchers developed HaDMC, a hybrid-action deep reinforcement learning framework that employs a policy learning algorithm to generate latent continuous actions. An action decoder was specifically designed and trained, incorporating two pipelines to convert the latent continuous actions into the original hybrid actions, enabling direct interaction between the drone/mobile charger and their environment. The framework implements a mutual learning scheme during model training that emphasizes collaboration over individual actions.Through extensive numerical experiments, the team evaluated HaDMC against state-of-the-art approaches, with results demonstrating the solution's effectiveness and efficiency.

The paper's first author is Dou Jizhe, a master's student at the School of Information Science and Technology (School of Artificial Intelligence), with Associate Professor Sun Guodong serving as corresponding author. This work was supported by the National Key R&D Program of China: "Long-Endurance UAV-Based Intelligent Monitoring of Forest Megafauna" (2022YFF1302700).

Paper link: https://ieeexplore.ieee.org/document/10925829

Written by Reporter Dou, Sun Guodong
Translated and edited by Song He
Reviewed by Yu Yangyang