Design of a forest fire patrol UAV system based on dual-sensor payload and YOLOv8n detection

Abstract

Forest fires pose severe threats to ecological environments and public safety due to their sudden onset and rapid spread. This paper presents a forest fire patrol and assisted identification system based on a quadrotor UAV platform, integrating a dual-sensor payload comprising a 30× optical zoom visible-light camera and a 640×512 infrared thermal imaging camera, with a lightweight YOLOv8n algorithm deployed on an onboard edge computing module. The system establishes a cooperative dual-channel alarm mechanism combining visible-light semantic detection, thermal imaging anomaly verification, and ground station-assisted review, which effectively reduces false alarms while maintaining sensitivity to early-stage fires. Taking Liuxihe National Forest Park in Guangzhou as the design scenario, key parameters are determined: maximum takeoff weight (MTOW) approximately 6.5 kg, cruising speed approximately 7 m/s, and mission endurance no less than 25 minutes. A ground station simulation system is developed using PyQt5 and OpenCV to verify patrol image display, suspected fire marking, UAV status monitoring, and alarm-assisted manual review function across three representative fire scenarios. Results demonstrate that the proposed integrated system achieves good task adaptability and potential engineering feasibility for grassroots forest fire monitoring applications.