Multi-UAV Energy-Efficient Wildfire Coverage Optimization

Uncrewed aerial vehicles (UAVs) are expected to play a pivotal role in 6G networks due to their versatility and adaptability. One potential application for UAVs is wildfire coverage, as they can carry various sensors, including cameras and antennas. This study focuses on the multi-UAV trajectory optimization for wildfire coverage while satisfying multiple constraints, including the UAV dynamics, network connectivity, and limited energy batteries. The resulting complex optimization problem is time-varying and non-convex. To address this challenge, reinforcement learning, specifically the twin-delayed deep deterministic policy gradient algorithm, is adopted. A distributed learning procedure is devised to allow parallelization and significant reduction of the training time. The result is high coverage at standard flying altitudes with finite energy batteries.