In-Depth Safety Analysis for Right-Turn Vehicle-Pedestrian Interaction Based on Checkpoint Video

This paper established a high-precision video detection method based on intersection cameras to analyze the traffic conflict between vehicles and pedestrians. The framework of the method is divided into automatic video trajectory extraction, conflict risk computation, and risk field analysis. A signal intersection at Lhasa, Tibet, was used to validate the method and extract conflict data for vehicles and pedestrians. A total of 1,213 conflicts data from 24 h were collected, and the conflict characteristics of pedestrians and vehicles at crosswalks in days and cycles were analyzed, respectively. Then how the risk field guides the work of the ADAS system was explained with a case study. The results show that the highest number of pedestrian-vehicle conflicts on the east side of crosswalks occurred in three time periods, which coincides with the peak periods for pedestrians. Risk field guides ADAS system effectively avoids higher-risk driving routes.