Considering cut-in behavior in mixed traffic: A longitudinal potential field-based car-following model for autonomous vehicles

Abstract

To assure safety and comfort for autonomous vehicles (AVs) in the mixed traffic environment, a longitudinal potential field-based car-following model considering the cut-in behavior from an adjacent vehicle is established in this study. The longitudinal potential field (LPF), which includes the cut-in potential field, the hazard potential field and the velocity potential field, is constructed to quantitatively characterize the influence of the surrounding environment on the AVs. The cut-in potential field is created to indicate the possibility of cut-in behavior from the adjacent vehicle, the hazard potential field is intended to depict the risk of colliding with the preceding vehicle, and the velocity potential field is created to show how velocity impacts the following behavior. The proposed model is obtained by calculating the field force in the LPF, and its parameters are calibrated using the genetic algorithm. Comparative simulations are executed to verify the remarkable performance of the proposed model, in terms of acceleration, velocity, and headway tracking as well as responding to the cut-in behavior under three different driving conditions.