Equilibrium Traffic Dynamics with Mixed Autonomous and Human-Driven Vehicles and Novel Traffic Management Policies: The Effects of Value-of-Time Compensation and Random Road Capacity

Abstract

Emerging autonomous vehicles (AVs) are expected to bring about a revolution in both the automotive industry and transportation systems. Introducing AVs into the existing mobility system with human-driven vehicles (HVs) yields mixed traffic with the following new features: in-vehicle compensation on value of time for AV users, distinct road capacities for pure AV and HV flows, and stochastic road capacity for the inseparable AV-HV traffic pattern. In this paper, we aim to investigate equilibrium traffic dynamics for the morning commuting problem where AVs and HVs coexist in a transportation corridor by considering these new features, and also explore several novel mixed AV-HV traffic management strategies. The AV-HV traffic pattern could be either separable (i.e., pure AV flow and pure HV flow depart from home in different periods) or inseparable, depending on the user profile condition. In addition to deriving departure time equilibriums for scenarios with separable traffic flows, significant effort is put into the scenario with an inseparable AV-HV traffic pattern, where stochastic road capacity is taken into account. Based on these equilibrium traffic analyses, we propose and explore some new traffic management strategies, including AV certificate of entitlement management scheme for scenarios with separable traffic flows and departure-period management (DPM) scheme and lane management policies for the scenario with an inseparable AV-HV traffic pattern. Eligibilities for applying these strategies are analytically derived and extensively discussed, and numerical experiments are conducted to demonstrate our theoretical findings and reveal the underlying impacts of road capacity randomness. Some lessons learned from the numerical experiments are (i) overlooking the impact of road capacity uncertainty will lead to an overestimation of system performance and even yield biased policymaking, (ii) the full dedicated-lane policy is the preferred option for the medium-level AV situation and partial dedicated-lane policies are more attractive choices for the early AV era or a market with a high AV share, and (iii) the DPM scheme could be a better substitute for partially dedicated-lane policies. Funding: This study was supported by the Ministry of Education of Singapore [Project T2EP40222-0002 under the MOE Tier 2 Grant] and the National Natural Science Foundation Council of China [Grant 72001133] and the Excellent Young Scientists Fund Program (Overseas). Supplemental Material: The online appendix is available at https://doi.org/10.1287/trsc.2021.0469 .