Vector Field based Tracking of a Dynamic Wildfire Boundary with an UAV

Abstract

Accurate information concerning the ongoing wildfire event is essential for effective and safe wildfire fighting. In this paper, we consider the problem of guiding a single Unmanned Air Vehicle (UAV) equipped with an onboard camera to track the dynamic wildfire boundary. As the field of view of the camera is limited, only partial wildfire can be sampled if the wildfire scale is large. After processing the sampling data, the coordinates of partial observed points on the wildfire boundary are available for computing the distance between the UAV and the wildfire boundary segment. On the basis of the distance, a vector field whose integral converges to the observed segment of wildfire boundary is constructed for designing the guidance law. In order to validate the performance of the proposed controller, we performed experiments with both synthesized and real data. Simulation results indicate that with partial observation, the proposed control strategy can guide the UAV to converge to and patrol along the advancing wildfire boundary.