Modeling Human-Like Driving Behavior at a Signal Intersection Based on Driver Risk Field Model

Autonomous intersection management systems aim to efficiently control connected and autonomous vehicles at urban intersections. However, current driving behavior models face challenges in accurately capturing the distinctive human driver characteristics specific to intersection interactions. This article introduces a human-like driving behavior model based on the driver's risk field (DRF) for intersection scenarios. The DRF represents the driver's belief regarding the likelihood of an event occurring, and the associated cost function is determined by the consequences of said event. A driving simulation experiment was conducted at a signalized intersection to evaluate the model, and the results were compared with a human-like driving behavior model. The results show that the proposed model has a high degree of fit. Furthermore, a statistical analysis of the data distribution demonstrates that the predictions generated by the driver model align closely with the driving behavior observed in the signalized intersection experiment.