3D Thermal Detection of Humans in Smoky Indoor Environments

Abstract

Human victim detection during search and rescue missions is a challenging task for firefighters. Especially in smoky indoor environments, firefighters are limited in their vision and orientation. Mobile robots have gained a significant attention in recent years, as they support firefighters during their mission. These robots are equipped with several sensors and operate autonomously or remote controlled by an operator outside the hazardous environment. In this paper, a new approach to detect human bodies in a light smoky indoor environment is proposed. The approach consists of a monocular RGB camera and a low resolution thermal camera $(8\times 8\ \text{px})$ mounted on a mobile robot. The mobile robot is controlled remotely by an operator outside the scene. A Large-Scale Direct Simultaneous Localization and Mapping method (LSD-SLAM) is used to generate a 3D thermal map of a human body in the smoky environment by superimposing thermal information onto a 3D surface reconstructed by a monocular RGB camera. Finally, the approach is approved by an experiment in a light smoky indoor environment.