Modeling decision-making process of drivers during yellow signal phase at intersections based on drift–diffusion model.

• Drivers' decision-making time and outcomes during the yellow light are predicted. • A drift–diffusion model was adopted to characterize the decision-making process. • We examine drivers' decision-making behavior by driving simulator. • The model is validated for individual and representative drivers. • The arrival time has a significant impact on driver decision response time. The decision-making behavior of drivers during the yellow signal phase has a significant impact on intersection safety. To analyze the decision-making process, we conducted surveys on driver behavior during yellow signal phase. A drift–diffusion model was established to analyze factors associated with driver decisions. The model can accurately predict driving decision outcomes (whether to proceed through the intersection during the yellow signal phase) and the decision-making times of different drivers. Driving data were collected using a driving simulator, including 15 participants in 210 tests in seven scenarios (3150 experimental samples). Drivers with similar driving behaviors were grouped. The model was validated using both in-sample and out-of-sample data for both individual and representative drivers. It was found that the error rate of the predicted data was approximately 7 %. Different arrival times had a significant impact on decision response time. Drivers tended to make faster decisions when the arrival time was less than 2 s due to the urgency of the decision. The findings can help understand the underlying cognitive mechanisms of driver behavior during the yellow signal phase. [ABSTRACT FROM AUTHOR]