Autonomous Wildfire Tracking Systems Based on UAV and Perspectives of Wildfire Digital Twin

Abstract

The ever-burgeoning development of digital twin systems lays the path for a new approach to wildfire monitoring and prevention. A real wildfire digital shadow reflects the fire development, providing a promising platform for firefighting decisions. However, currently, static terrain information cannot be integrated with dynamic wildfire locations, which is a great hindrance to the utilization of digital twin systems. The flame trend fails to be mapped in the digital shadow. The key to solving this issue is to ensure real-time data transfer. Thus, UAVs equipped with vision systems could be leveraged to maintain surveillance, transferring the real-time information to the digital twin systems. Initially, the study reviews the UAVs in wildfires. Specifically, the history of UAV tracking is introduced. Furthermore, the autonomous wildfire tracking system is proposed to assist in locating the fire front. A YOLO-based algorithm is utilized to detect the fire front. The UAVs use an orthographic view to locate the wildfire, controlling and maintaining the gesture with a perspective lock on the flame front. Then, UAVs track and follow the flame front without human intervention. In addition, the combination of UAV location information based on GPS and camera view information depicts the wildfire in the digital twin systems. This study aims to act as a pre-processing step for wildfire digital twin data transfer. It will provide a basis for the precise wildfire portrait in the virtual systems.