Real-time risk assessment model for multi-vehicle interaction of connected and autonomous vehicles in weaving area based on risk potential field

To assess the operational risk of multi-vehicle interactions for connected and autonomous vehicles (CAVs) in the weaving area, this study defines the ‘multi-vehicle interaction risk potential field’ based on the similarity between vehicle risk and potential field, on the basis of which the risk potential field between the interactive CAVs and ego CAV is determined. Then, a real-time risk assessment model of CAVs in the weaving area is developed, and a surrogate safety measure R of the CAVs’ operational risk is proposed. Subsequently, the kinematic data of CAVs are obtained by simulation. Furthermore, the genetic algorithm is used to calibrate the parameters of the real-time risk assessment model. The validity of the model is demonstrated by comparing the R value and traffic conflict indicator (TTCi) for the car-following and cut-in behaviour in the weaving area scenario. The validation results of typical cases show that the variation ranges of R in the car-following and cut-in scenarios are 2.2 and 1.9 times those of TTCi, respectively, which can more effectively denote the operational risk of CAVs during multi-vehicle interactions in weaving area. The study can be used to provide reference for CAVs’ driving decisions in weaving areas and promote their safety in multi-vehicle interaction scenarios.