Author(s): Yu Zhe; Lu Wenjing

Linked Author(s):

Keywords: Deep reinforcement learning; Drone swarm; Flood emergency monitoring

Abstract: In recent years, drone remote sensing technology has been widely employed in flood prevention and mitigation. The critical point of flood emergency monitoring lies in obtaining timely the leakage warning information of dams. Flood emergency monitoring is often confronted with a multitude of challenges, encompassing severe climatic conditions, intricate terrain, restricted monitoring timeframes, and stringent data accuracy requisites. Current mainstream drone remote sensing monitoring methodologies, characterized by their reliance on predetermined flight trajectories, encounter a plethora of challenges, including incomplete dam image shooting, insufficient image resolution, delayed emergency response, and lengthy data acquisition times. To address these challenges, we propose an drone swarm remote sensing monitoring method based on deep reinforcement learning. Drone swarm can autonomously adjust the drones' shooting positions in order to improve the integrity and resolution of dam images, while reducing the shooting time in challenging climatic conditions and complex environments. The versatility, low cost, and high robustness of drone swarms offer significant advantages in addressing time-sensitive flood emergency monitoring, and represent a key area for future development.

DOI: https://doi.org/10.3850/iahr-hic2483430201-291

Year: 2024