UAV Trajectory Optimized for Urban Inspection Using Communication-Aware Maps.

This research addresses the critical challenge of maintaining reliable communication for Unmanned Aerial Vehicles (UAVs) during urban inspection tasks, where signal quality can vary significantly. The proposed framework focuses on optimizing UAV trajectories by making them "communication-aware." It achieves this by first constructing a Channel Knowledge Map (CKM) using a diffusion model, which accurately reconstructs high-fidelity global channel quality distributions from sparse observation data, minimizing flight overhead. Building upon this CKM, the framework employs a sophisticated global-to-local graph attention network combined with a soft actor-critic algorithm. The graph attention network is designed to solve complex combinatorial problems, generating an optimal and communication-aware sequence for inspecting targets. Subsequently, the soft actor-critic algorithm refines the continuous flight path, ensuring smoothness and dynamically avoiding areas with poor communication. Simulation results confirm that this method effectively guides UAVs through high-quality channel regions without needing real-time feedback, significantly enhancing both trajectory efficiency and communication reliability.