Single-variable scenario analysis of vehicle-pedestrian potential crash based on video analysis results of large-scale naturalistic driving data

Abstract

Vehicle-pedestrian crashes are big concerns in transportation safety, and it is important to study the vehicle-pedestrian crash scenarios in order to facilitate the development and evaluation of pedestrian crash mitigation systems. Many researchers have tried to investigate the pedestrian crash scenarios relying on crash databases or pedestrian behavior prediction models, both of which have some limitations like limited generalizability of the results, missing of important information, biased results. In this study, we propose to study the potential crash scenarios as one surrogate targets of the actual pedestrian crash scenarios. Extended from several previous studies, one single-variable scenario analysis is completed based on the video analysis results of one large-scale naturalistic driving data collection focusing on recording pedestrian behaviors in all kinds of situations. Through calculating potential conflict rates and applying chi-square tests for around 40 attributes from 12 scenario variables individually, this study has found out that number of pedestrians, pedestrian moving speed, pedestrian moving direction, vehicle moving direction, road type, road location, and existence of road separator/median are all important scenario variables for potential pedestrian-vehicle crashes.