查看文章

In the U.S., in 2015 alone, there were approximately 35,000 fatalities and 2.4 million injuries caused by an estimated 6.3 million traffic accidents. In the future, it is speculated that automated systems will help to avoid or decrease the number and severity of accidents. However, before such a time, a broad range of vehicles, from non-autonomous to fully-autonomous, will share the road. Hence, measures need to be put in place to improve both safety and efficiency, while not compromising the advantages of autonomous driving technology. In this study, a Bayesian network model is developed to predict the severity of an accident, should it occur, given the road and the immediate environment conditions. The model is calibrated for the case of traditional vehicles using pre-crash information on driver behaviour, road surface conditions, weather and lighting conditions, among other variables, to predict two …