Identifying Major Accident Scenarios in Intersection and Evaluation of Collision Warning System

Abstract

To improve the safety performance of the vehicle, various Advanced Driver Assistance Systems (ADAS) have been developed and apply to the vehicle. Among various ADASs, the collision warning system is one of the representative technology and contributed to prevent the accident and mitigate the severity of the accident. However, existing collision warning systems are designed to prevent the forward collision in an uninterrupted flow, so the safety performance in the interrupted flow near intersection have not been intensively studied. In this paper, the safety performance of Camera-Based collision warning system (CBS) and Radar-Based collision warning system (RBS) are evaluated near the intersection. For this, we newly derived the sixteen vehicle-to-vehicle accident scenarios near the intersection with a crash and near-crash data according to the movements of the subject vehicle and the opponent vehicle. Scenarios derived in this paper differs from the derived scenarios in the previous researches by dealing with more various movements of the subject vehicle such as turning movement. With the derived sixteen vehicle-to-vehicle accident scenarios, we evaluate how effectively CBS and RBS work in the intersection accident scenarios. The results show that CBS and RBS cannot prevent the all accident scenarios due to the limited field-of-view. With the 0.7 accident prevention rate, CBS can prevent the accident in five scenarios, and RBS can prevent the accident in four accident scenarios. To prevent all accident scenarios, another system that can detect the vehicle movement with longer range than CBS and RBS is needed. In the future study, communication-based collision warning system will be developed to further improve the safety performance of collision warning system.