Advancing Off-Road Operations: Comparative Assessment of Object Detection Methodologies

Off-road operations such as search and rescue or goods delivery in remote wilderness environments demand efficient and effective strategies to locate individuals. Traditional Search and Rescue (SAR) methods, such as canine units and human-assisted search, face challenges in timeliness, cost, scalability and adaptability. To address these limitations, we explore the integration of ground robots equipped with LiDARs into SAR workflows, focusing on off-road scenarios where adverse weather or complex terrain hinder conventional methods like Unmanned Aerial Vehicles (UAVs). Our study conducts a comprehensive analysis of single-sensor LiDAR algorithms for small object detection in off-road environments, utilizing a newly curated dataset tailored for SAR applications. We identify the critical role of sensor configuration and environmental factors in detection performance, providing insights for designing robust detection systems. By filling gaps in existing literature and elucidating the strengths of single-sensor LiDAR algorithms, we contribute to advancing off-road operations methodologies and technology. Our findings underscore the potential of integrating ground robots into off-road operations and offer valuable guidance for enhancing search capabilities in challenging environments.