Rescuebot-Thermal Imaging for Human Detection

Abstract

Natural disasters pose immense challenges, particularly when it comes to swift identification and rescue of victims, which is critical to saving lives. Unfortunately, existing search and rescue technologies often fail, struggling with issues such as low visibility, the absence of real-time tracking, and the difficulty of distinguishing victims from surrounding debris. To address these challenges, we present the Rescuebot, an innovative hexapod robot that takes advantage of advanced thermal imaging technology and deep learning algorithms to improve victim detection during rescue operations. The Rescuebot is designed to navigate through environments where visibility is severely compromised, such as areas filled with dense rubble or heavy smoke. A standout feature of this robot is its integration of the YOLO (You Only Look Once) object detection model, specifically the latest version, YOLOv8. This model has been trained on a rich thermal imaging data set sourced from Roboflow, allowing it to accurately identify human victims in a variety of complex scenarios. By enhancing the capabilities of search and rescue missions, the Rescuebot aims to make a meaningful impact in critical moments after a disaster. Our project combines software and hardware integration using a Raspberry Pi for autonomous victim detection and environmental mapping. Extensive testing in simulated environments optimizes the Rescuebot's performance for extreme conditions, including high temperatures and low light. In addition, the system features real-time tracking, enabling swift responses during rescue operations. By overcoming existing technological limitations, the Rescuebot significantly advances disaster management, with applications extending to industrial safety and military operations, ultimately aiming to save lives in emergencies.