Enhancing Right-Turn Safety at Intersections: AI-Based Intelligent System Development and Driving Simulator Testing

Le résumé fourni par la source

A recent revision of the Korean Road Traffic Act requires drivers to stop not only for pedestrians already crossing at intersections but also for those about to cross, significantly altering right-turn behavior and increasing driver responsibility. This change has introduced confusion, particularly at intersections with concurrent pedestrian phasing. In this study, a detection system was developed that can: 1) determine whether a driver is able to make a right turn; and 2) identify illegal right turns. Based on this system, a device was designed to provide right-turn information, and the effectiveness of the installation was evaluated using a driving simulator similar to real road conditions. Based on this system, a device was designed to provide right-turn information, and the effectiveness of the installation was evaluated using a driving simulator similar to real road conditions. An algorithm was developed using CCTV video based on YOLOv8, and the system was shown to be highly accurate. In addition, driving experiments were conducted to analyze the resulting data using various driving-related indicators, including vehicle speed, deceleration, and brake pedal force. The results showed that when right-turn information was provided, drivers exhibited reduced speeds (by approximately 2–4 km/h), increased deceleration, and earlier and stronger brake responses. The findings have important implications for traffic safety policy making and highlight the critical role of automated enforcement systems in mitigating vehicle–pedestrian crashes and creating safer urban environments.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.