A new risk estimation model of bayesian network for adapting to driving environment changing

In recent years, research on automated driving of automobiles is being promoted, and accidents caused by human error by driving support systems are also expected to decrease. However, most of the accidents occur because the risk that the driver feels subjectively is too small. Therefore, to reduce the number of traffic accidents, it is necessary to raise danger perception while driving. There are two kinds of risk in the driving environment: the subjective risk felt by the driver and the objective risk existing in the driving environment. In this research, we construct a model to estimate each risk value by using two pieces of information: traffic environment information obtained from the front image of the vehicle and driving operation information of the driver. Furthermore, by combining them the risk of adapting to the driving environment is determined, and acts to raise drivers’ perception of danger.