An Improved Car-Following Model with Consideration of Driver's Overtaking Intention

The analysis and modeling of car-following behavior is an indispensable part in the research of intelligent transportation research. And the car-following model can provide technical support for automatic driving control. However, the existing models rarely consider the driver’s driving style. To tackle this, this paper proposes a model which reflects the real driving behavior more accurately. Specifically, roadside LiDAR is used to obtain high-resolution vehicle trajectories data from real traffic scenes and it turned out that the driver’s overtaking intention have a great impact on the following behavior. Based this, this paper proposes an improved car-following model with consideration of driver’s overtaking intention. In addition, the performance of the proposed model is analyzed and evaluated with real-scene data. Extensive experiments reveal that the proposed model can reflect driver’s behavior characteristics in real traffic flow effectively, and is better than GM model and Newell model in terms of the prediction accuracy. The study results are of great significance to reveal the evolution law of traffic flow. Furthermore, this research is conducive to improving the safety and comfort of autonomous driving in the future.