A review of traffic behavior and intelligent driving at roundabouts based on microscopic perspective

Abstract

Abstract The contradiction between the increasing traffic and the relatively poor roundabout infrastructure is getting stronger. The control and optimization of the macroscopic traffic flow need to be improved to resolve congestion and safety problems at roundabouts and the connected road network. In order to better understand the gaps and trends in this field, we have systematically reviewed the main research and developments in traffic phenomena, driving behavior, autonomous vehicles (AVs), intelligent connected vehicles, and real vehicle trajectory data sets at roundabouts. The study is based on 388 papers about roundabouts, selected through a comprehensive literature search. The review demonstrates that based on a microscopic perspective, sensing, prediction, decision-making, planning, and control aspects of AVs and intelligent connected vehicles can be designed and optimized to fundamentally and significantly improve traffic capacity and driving safety at roundabouts. However, the generation mechanism of traffic conflicts among traffic participants at roundabouts is complex, which is a tremendous challenge for the systematic design of AVs. Therefore, based on naturalistic driving data and machine learning theory, it is an important research direction to build driver models by learning and imitating human driver decision-making and driving behaviors.