GIPUT: Maximizing Photo Coverage Efficiency for UAV Trajectory

Shaoting Feng, Qinya Li, Yaodong Yang, Fan Wu, Guihai Chen

Abstract: Unmanned Aerial Vehicles (UAVs) used for photo coverage require trajectories that differ significantly from traditional solutions. Traditional trajectories often treat objects as mere points, neglecting their geometric properties. In contrast, UAV trajectories must consider varying camera angles and viewpoints to capture comprehensive images. In this study, we propose a novel UAV trajectory algorithm that maximizes photo coverage efficiency by integrating computational geometry and deep reinforcement learning. Our approach accurately models objects as realistic shapes, enabling precise computation of UAV photo coverage. This model empowers the UAV to learn optimal trajectories considering multiple factors, including photo coverage, energy consumption, and bandwidth utilization. Extensive experiments were conducted to evaluate our proposed model across diverse scenarios. The results demonstrate a significant enhancement in UAV trajectory efficiency for aerial photo coverage compared to existing state-of-the-art approaches. This improvement facilitates a more effective image capture process, leading to broader coverage and reduced energy and bandwidth consumption.