Sistema de localización de personas en catástrofes utilizando vehículos aéreos no tripulados (UAV’S)

Abstract

The advancement of new technology for research and rescue systems has developed significantly in our country. Therefore, the objective of this investigation was to design a system which allows the detection and location of individuals in a case of a catrostofy, through the integration of Unmanned Aerial Vehicles (UAV’s), a thermal camera and another RGB. To incorporate the thermal camera, to the UAV, a support was designed to allow the viewing angle to be adjusted and four supports printed in thermoplastic polyurethane were created to protect it during takeoffs and landings. Once it begins functioning, the capture of the thermal signature of human radiation begins and its transmission to the ground station, where the video is processed in real time to obtain the graphic interface to allow the visualization of the thermal camera and RGB camera along with the data location. The UAV was tested on different heights, times and locations to simulate realistic scenarios, generating more than 1000 images used to train a neural network in YOLOv8l with an accuracy of 90%. After various tests of the system, it was determined the optimal flight height for detecting a person is 16 meters, along with an inclination of the thermal camera less than 45 degrees to avoid reflectivity in the field of view.