Developing the guidelines for managing autonomous vehicle flows on congested highways: A case study of MD-100

Abstract

Recognizing the need for responsible highway agencies to effectively manage emerging autonomous vehicles (AV) flows in contending with daily recurrent congestion, this study presents a systematic procedure for understanding the impacts of AV flows on traffic conditions under different AV behavioral mechanisms (i.e., car-following and lane-changing), and different penetration rates. Research results show that the presence of AV flows, depending on their adopted behavioral mechanisms, may have significant (either positive or negative) impacts on the overall traffic conditions. Hence, it is essential for responsible highway agencies to have proper operational guidelines to manage and coordinate AV flows. To demonstrate the proposed methodology, this study has carried out extensive simulation experiments using a congested segment of the MD-100 network (a multilane highway segment located in Maryland) under various AV penetration rates and observable behavioral patterns. The collected Measures of Effectiveness highlight that at each AV penetration level there exists a set of optimal behavioral patterns for the AV flows to coordinate with non-AV flows via the Vehicle to Infrastructure or Vehicle to Vehicle infrastructure so as to maximize the roadway capacity and minimize the resulting highway congestion.