Modular unmanned ground vehicle architecture for search and rescue operations: autonomous SLAM and AI-driven thermal victim detection

Purpose The catastrophic effects of natural disasters on human life and the critical importance of timely response have significantly increased the demand for autonomous search and rescue (SAR) technologies such as unmanned ground vehicles (UGVs) and unmanned aerial vehicles (UAVs). The capability of these systems to autonomously navigate hazardous and inaccessible zones and perform real-time human detection is critical for enhancing operational efficiency and ensuring the safety of rescue teams. To meet these requirements, this study aims to present the design and development of a modular, structurally robust UGV for post-disaster autonomous SAR operations. Design/methodology/approach The proposed platform integrates an RGB-depth based visual Simultaneous Localization and Mapping (v-SLAM) pipeline on a high-performance NVIDIA Jetson Orin Nano Developer Kit. The sensing system includes an Intel RealSense D435i depth camera and a thermal imaging module for detecting human heat signatures. The chassis is 3D-printed using PETG material for durability in harsh environments. Mobility in unstructured environments is supported by a Passive Convertible Single (PaTS) design. For real-time human detection, a You Only Look Once (YOLO) model was deployed on thermal images acquired by UGV. Findings Tests show that proposed UGV system achieved a precision of 97.75%, mean average precision (mAP) of 97.21% (IoU = 0.50) and a recall (sensitivity) of 93.33% in thermal human detection and featuring sub-10-second UGV-to-handheld hot-swaps. Originality/value Therefore, the proposed study provides an autonomous and robust UGV for SAR operations by integrating real-time human detection and v-SLAM-based navigation. This combination improves operational performance under limited visibility and accessibility conditions. In addition, the system’s ability to be converted into a handheld device provides additional operational flexibility.