Analysis and modeling of driver behavior on pedestrian crossing situations (2nd report: Analysis of a crossing diagonally situation using JARI-ARV)

Pedestrian-crossing-related accidents were 26.0% of all fatal accidents that occurred in Japan in 2012. To prevent pedestrian-versus-vehicle accidents, several driver-support systems (alarms and automatic brakes) have recently been commercialized. For verification of the validity of these driver support systems, it is important to understand driver behavior regarding pedestrian crossing. In this study, we examined driving performance for situations where pedestrians suddenly crossed the road to construct a driving behavior model. Experiments were implemented using an instrumented vehicle, JARI-ARV, which was able to reproduce critical situations where pedestrians suddenly crossed the road in front of the instrumented vehicle using augmented reality technologies (computer graphics). Several walking patterns including diagonal crossing of the road were set up as experiment conditions. Consequently, a collision avoidance behavior model for a particular complex crossing pattern was determined.