|dc.description.abstract||Driving safety has been accorded increased attention in recent years. However, the casualty caused by traffic accidents has continuously been increasing over the years. Traffic accidents can be attributed to reasons related to “driver, vehicle, and the road.” In the case of conflict between these three factors, the driver’s avoidance action may (or may not) cause accidents. Driving performance also relates to the ability to avoid traffic accidents. Against this backdrop, this study will explore and measure driving performance in order to assess the crash risks. We developed a virtual scenario through a driving simulator, explored the assessment method of crash risks, examined the influential factors and degree of attribution by applying the study method, and provided the reference basis for the improvement of traffic safety through the prediction model of crash risks.
This study established comprehensive indexes of driving performance based on 50 subjects who recruited the test and compared the influence of gender (male, female) and auditory alarm contents (null, beep sound, speech message) on driving performance. The experimental result indicates the interactive influence between gender and the alarm system on the standard deviation of lane deviation. We developed the comprehensive indexes through principal component analysis method. The study results indicate that the risk of the driver is bigger when the fraction of the principal component is higher, and the risk of operation is higher when the time of perception is longer, the average speed is higher, the standard deviation of speed is smaller, the average lane deviation is higher, and the standard offset of lane deviation is bigger.
This study investigated the causality between driving performance and accident risks through 30 subjects who recruited the test in order to explore the influence of intersections of young drivers and traffic factors (intersection collision warning system, location of the accident intersection, days of driving per week) on driving performance and crash. The study employed path analysis to explore the causality between driving performance and accident risks. The result indicates that the alarm system has no direct influence on crash risks, and the probability of crash can be reduced by improving driving performance. Moreover, the location of intersection accident is related to experiences of driving accidents. These accident experiences can reduce the risks of subsequent crash, and they have a direct influence on both driving performance and crash. Thus, risks of crash can be reduced by improving driving performance. Moreover, driving experience, that is, days of driving per day, has a direct influence on crash but not on driving performance.
This study investigated studied the predication of crash risks by examining the driving performance of 28 subjects who recruited the test. We also compared the influence of the location of traffic accidents (roadway segment, intersection) and auditory alarm contents (null, beep sound, speech message) on driving performance and crash. The result indicates that different locations of traffic accidents have different demands for information source. Proper acoustic warning allows driver to avoid crash through proper actions. We conducted logistic regression analysis on the prediction of crash risks and found that the perception-reaction time has the largest influence on crash.
The findings indicate that building comprehensive indexes with principal component analysis, deciding the causality with path analysis, and predicting crash by logistic regression analysis will help us understand the relationship between driving performance and crash risks. This can be expanded to other topics for study in the future in order to improve traffic safety further.