Evaluation of driver’s judgment assistance system at signalized intersection using incident database

Abstract

This study evaluates the effectiveness of a driver assistance system at signalized intersections by using an incident database. The research target is in the incident scenes of a vehicle and crossing pedestrians in which the vehicle proceeds straight to the signalized intersection. The evaluation of the effectiveness for the driver assistance system has three steps. Firstly, incident scenes for the crossing pedestrian during an amber traffic signal are extracted from the incident database by visual observation. Next, the extracted incident scenes are classified according to the behaviors of both the vehicle and the crossing pedestrian, and the location of the incident. Finally, assuming the driver follows the driver assistance system, the extracted incident scenes are analyzed as to what extent the incident could be prevented based on numerical simulation where the vehicle moves following the decision of the system whether continuing forward or stopping. For cases that the driver tries to enter the intersection despite the amber signal and being able to stop by the stop line, the driver assistance system is supposed to prevent the incident effectively. For cases that the crossing pedestrian does not obey the pedestrian traffic signal, the driver assistance system is found not to be effective. However, if both the driver and crossing pedestrian obey their respective traffic signals, the driver assistance system can prevent the incident effectively in many cases.

Keywords

• driver assistance system
• signalized intersection
• amber traffic signal
• dilemma zone
• drive recorder
• pedestrian
• incident database