Abstract

In this research, a UAV (unmanned aerial vehicle) system for detecting and locating forest fires is developed. This system uses the information from two cameras: one visual camera and one thermal camera for fire detection and navigation. Two images from these cameras are aligned before they can be used. An alignment process is created and a homography matrix is computed so that it is assured that the two images are aligned and every pixel is the same in both images. By combining data extracted from these cameras, it can verify whether there is no fire or a real fire is happened or a fake fire. A two-degree-of-freedom (2DOF) frame is implemented for testing the tracking and locating capabilities of the UAV. Both cameras are mounted in this frame. This system is based on the concept of image-based thermo-visual servoing (IBTVS). That is, it extracts thermal and visual information of a scene and then it computes the position of a desired object (fire) and commands the servos to move to the desired position. This frame can simulate 2DOF movement of a UAV and can be used to test the developed fire detection algorithms. Finally, a co-simulation system is presented to verify the application of fire detection in a simulated UAV. The experimental tests demonstrate that the developed algorithms can guide the UAV to fly on a predefined path and look for any possible fire. As soon as a fire is detected, the system will alarm and calculate the location of fire and fly the UAV over the fire for further investigations.