Proactive Assessment of Pedestrian and Bicycle Safety at Intersections Through Video Analysis of Right Turn on Red (RTOR) Maneuvers

Right-turn-on-red (RTOR) maneuvers increase traffic efficiency but create conflicts with pedestrians and cyclists, raising safety concerns that crash records often fail to capture. This study develops a proactive framework using video analytics and surrogate safety measures to evaluate RTOR risk. One week of video data from the Intelligent Transportation Management Center (ITMC) in Chula Vista, California, was analyzed with fine-tuned object detection models (YOLOv11 and RT-DETR). Post-Encroachment Time (PET) was computed to classify conflict severity using a cumulative distribution function (CDF)-based approach. We identified 65 RTOR events, including eight critical pedestrian or cyclist conflicts. YOLOv11 outperformed RT-DETR in detection accuracy and speed. Results showed critical events were linked to higher approach speeds, lower PET values, and greater traffic density. This framework enables proactive safety evaluation and supports before–after studies of countermeasures to reduce critical events.