Modelling the dual dynamic traffic flow evolution with information perception differences between human-driven vehicles and connected autonomous vehicles

Abstract

The introduction of connected autonomous vehicles (CAVs) potentially improves the link capacity and backward wave speed of traffic flow, while the advanced communication technology could well make it possible to allow CAV users to share their travel information. To bridge the knowledge gaps in the network evolution under mixed environment of human-driven vehicles (HVs) and CAVs, it is essential to explore multi-dimensional dynamic traffic assignment. An inertia-based multi-class dual dynamic traffic assignment (IMDDTA) model is proposed to capture the intraday and diurnal variations of the mixed traffic flow under the disequilibrium state simultaneously. Specifically, in this study we consider the inertia of HV users as well the information-sharing behaviour of CAV users respectively, characterized by different extensions of the multinomial logit (MNL) model. To demonstrate the properties of the model, two numerical case studies are conducted based on the Braess network and the Sioux Falls network. The results indicate an acceptable validity and applicability of the model and provide valuable insights on the evolution of traffic flow under mixed environment.