Automated safety evaluation of intersections using advanced video recognition technology

The present study aims to assess the utility of advanced video recognition technology in assessing road safety at intersections. In particular, this study assessed safety performance of three complex intersections in Brisbane by using safety surrogates automatically measured by an advanced video recognition technology. Traffic movement data on typical weekdays at the chosen intersections were video recorded. Conflict analysis was then performed by analyzing traffic interactions among vehicles, pedestrians and cyclists. In particular, Time-To-Collision (TTC) and Post Encroachment Time (PET) were used to identify frequency, severity and locations of the conflicts at each intersection. Subsequently, suitable countermeasures were identified to reduce these conflicts and improve safety at these intersections