Optimization of Lane-Changing Decision Model for Enhancing Human Factors Based on Prospect Theory

Lane-changing is a basic driving behavior, but current modeling approaches fail to portray complex lane-changing processes. Furthermore, most microscopic traffic models assume the rationality of drivers. Therefore, a discretionary lane-changing decision model based on prospect theory is developed, taking into account the driver’s limited rationality characteristics. First, the driver’s lane-changing decision factors are assessed using SHAP. Then, the prospect value is obtained from a combination of the driver’s weighing of objective probabilities and the value function, according to lane-changing decision factors. Then a binary logit model based on the prospect theory is developed to obtain the probability of lane-changing. It is found that the model not only has better accuracy and broader applicability but also reflects the limited rationality of drivers. The driver’s decision-making factors sorted by SHAP value more accurately reflect the influence of each factor on the drivers in the real scene.