Identification and Quantitative Analysis Technique of Road Obstacle Safety Risks

<div class="section abstract"><div class="htmlview paragraph">The transition from manual to autonomous driving introduces new safety challenges, with road obstacles emerging as a prominent threat to driving safety. However, existing research primarily focuses on vehicle-to-vehicle risk assessment, often overlooking the significant risks posed by static or dynamic road obstacles. In this context, developing a system capable of real-time monitoring of road conditions, accurately identifying obstacle positions and characteristics, and assessing their associated risk levels is crucial. To address these gaps, this study proposes a comprehensive process for rapid obstacle identification and risk quantification, composed of three main components: road obstacle event detection and feature extraction, risk quantification and level assessment, and output of warning information and countermeasures. First, a rapid detection method suited for highway scenarios is proposed based on the YOLOv5 model, enabling fast detection and classification of obstacles in highway environments. Second, a customized risk assessment model tailored to highway scenarios is developed using potential field theory, considering multiple influencing factors, including obstacle type, location, and road attributes. The proposed model provides a complete process for rapid obstacle identification and quantitative risk assessment. This system not only allows for early detection of potential hazards but also timely issuance of warnings, enabling intelligent connected vehicles (ICVs) to perform appropriate evasive maneuvers. This real-time decision-making enhances the safety of autonomous driving. These findings make a significant contribution to the development of intelligent warning systems and strengthen the deployment of ICVs in highway scenarios, supporting a safer and more reliable transportation system.</div></div>