Abstract
As part of efforts to help mitigate pedestrian-related accidents, driver support systems are being developed that detect pedestrians and warn the driver when the probability of a collision is high. Although the effectiveness of a system can be enhanced to some degree by verifying its activation performance on a test course, its real-world accident mitigation benefit cannot be predicted before the system is released. Therefore, a method has been developed that estimates the potential benefit of active safety systems by reproducing pedestrian accident scenarios that occurred over one year in Germany using the German In-Depth Accident Study (GIDAS) database, fixed-point observations in actual traffic, driving recorders, and tests using a driving simulator and actual vehicles. The same scenarios are then simulated considering activation of the active safety system. Following this method, the potential benefit of the active safety system can be estimated by comparing the differences with and without system activation. The scope of this method can be expanded to estimate the potential benefit of more complicated systems and it should prove to be an effective tool in the development of new active safety systems. Specifically, in this research, a method was developed that estimates the potential benefit of active safety systems designed to help mitigate accidents involving pedestrians crossing a road (i.e., crossing-pedestrian scenarios). The behavior of both vehicles and pedestrians was analyzed based on actual traffic accident data from the GIDAS database. Furthermore, nationwide statistics were analyzed to obtain the injury risk for drivers and pedestrians. It was found that the main factors affecting injury risk are the vehicle speed and the age of the pedestrian. Determining the injury probability in this way enabled the estimation of the number of slightly, seriously, and fatally injured people in each accident. After reproducing the accidents, the sum of the casualties in those accidents was defined as an estimate of the total number of casualties in one year in Germany. The trends of the reproduced accidents matched those of real-world accidents, and other parameters (such as vehicle and pedestrian speed distributions) also showed similar patterns between the reproduced and real-world accidents. These results verified that the estimation method reproduced the intended accident scenarios correctly. Crossing-pedestrian accident scenarios were reproduced based on data from nationwide statistics, direct traffic observations, driving recorders, and tests using a driving simulator and actual vehicles to create a comprehensive and accurate reproduction of traffic accidents involving pedestrians crossing a road. The relationship between impact speed and injury probability for drivers was also calculated. The casualties estimated by the accident reproduction matched the national statistics. Reproducing these accident scenarios enabled the effect of the safety system to be isolated, thereby identifying the extent and scope of the potential benefit for each system specification under development.
KEYWORDS – benefit estimation, crossing pedestrian