Modeling and Analysis of Microscopic Risk Avoidance Behavior of Homogeneous Driver Groups Under Risk Scenarios

To cope with the increasing number of road traffic crashes, it is critical to develop a driver assistance system that can provide early warning for vehicle collisions and control the vehicle at critical moments. However, to achieve this function, the driver assistance system must proactively understand drivers’ preferences and predict their risk avoidance behavior in risk scenarios, an area that currently lacks sufficient research. To address this issue, this study proposes a method for modeling microscopic risk avoidance behavior for homogeneous groups of drivers. Firstly, the risk field theory is established to achieve the basic driving risk assessment. Subsequently, a macro–micro collision-tendency probability calculation model is constructed to correct the basic driving risk values and obtain more accurate risk assessment results. Finally, a risk avoidance behavior model is developed by combining drivers’ risk response behavior and the psychology of desired speed pursuit. This study uses natural driving data for model validation. The results imply that the risk assessment indicator proposed in this study can reflect the driving risk under different risk phases. The risk avoidance behavior model accurately identifies vehicle acceleration fluctuations and matches drivers’ avoidance motivation in risk scenarios. In addition, the model parameters calibration results reveal significant differences among different driving groups; for example, risk perception and desired speed. This study aims to deepen researchers’ understanding of drivers’ risk avoidance behavior for designing driver assistance systems and road safety management.