Mixed Traffic Network with Semi-Autonomous Vehicles: A Three-Dimensional Macroscopic Fundamental Diagram and Its Application to Dual-Mode Perimeter Control

Abstract

The urban mobility landscape is evolving from traditional vehicles (TVs) to a mix including semi-autonomous vehicles (semi-AVs), which provide limited automation yet still require human oversight. This represents an important transitional phase in vehicle automation, often overlooked in studies favoring a binary division between TVs and fully-autonomous vehicles (fully-AVs). Our study addresses this gap by focusing on the more immediate, real-world mixed traffic of TVs and semi-AVs, where semi-AVs can switch between auto-driven and human-driven modes. This presents a complex scenario where human-driven vehicles (HDVs), comprising TVs and human-driven semi-AVs (HD-SAVs), coexist with auto-driven semi-AVs (AD-SAVs). To address this, we explore the use of dedicated lanes and develop a three-dimensional macroscopic fundamental diagram (3D-MFD) to analyze flow dynamics within mixed traffic networks of HDVs and AD-SAVs. Our findings identify the optimum proportion of HDVs and AD-SAVs (OPH\&A) that maximizes network flow at various congestion levels, highlighting the potential for enhancing network performance by adjusting the driving mode proportion. Additionally, we introduce and analytically derive two key concepts: the minimum penetration rate of controllable semi-AVs (MPR-CA) necessary to achieve OPH\&A, and the restricted OPH\&A for situations where MPR-CA is unmet. Leveraging these insights, we propose a novel dual-mode perimeter control strategy to regulate the mode choice of semi-AVs and optimize network performance. Simulation results confirm the efficacy of this strategy. Overall, our focus on the dual-mode capability of semi-AVs highlights an important, untapped aspect of the transition towards autonomous transportation. By applying the 3D-MFD and developing an effective control strategy for semi-AVs, our study sets a foundation for future traffic optimization efforts that utilize the nuanced control of semi-AV driving modes.

Keywords: Mixed traffic, three-dimensional macroscopic fundamental diagram, semi-autonomous vehicles, dedicated lanes, dual-mode perimeter control