Understanding and Harnessing Traffic Fundamental Diagram in the Era of Connected Automated Vehicles

点击翻译按钮获取中文摘要

The traffic ‘fundamental diagram’ describes fundamental relationships among the key traffic variables of flow, density, and speed. It is used to understand traffic phenomena and develop traffic management strategies. The adoption of connected automated vehicles is expected to bring systematic changes to traffic flow patterns and increase the complexity of traffic dynamics. This requires a deeper understanding of how stochastic and non-static traffic flow properties are manifested in the fundamental diagram. To this end, this project aims to develop a new theoretical framework to enhance the fundamental diagram with stochastic and non-static dimensions. The framework will be empirically grounded and developed based on trajectory data of human-driven vehicles and vehicles with automation features (adaptive cruise control). The framework will also utilize computational models to characterize the stochastic behavior of these vehicles, particularly when the exact algorithm of adaptive cruise control is unknown. The anticipated outcome of this project includes a better understanding of and models to describe stochastic and non-static features of highly heterogeneous traffic flow. This project will also engage a range of educational, outreach, and dissemination activities, including (1) industry and transportation agency outreach, (2) local K-12 school outreach, (3) Engineering Expo participation, (4) development of educational modules using the developed simulation tools, (5) promoting diversities by recruiting students from underrepresented groups.The traditional fundamental diagram is a time-invariant and deterministic description for traffic conditions in steady states. Further, it represents the collective behavior of vehicles, often missing a connection to vehicle-level behavior important for some dynamic traffic phenomena. Thus, the traditional fundamental diagram is not well equipped to describe complex traffic dynamics in highly heterogeneous traffic. This research will: (1) construct a generic stochastic calibration framework based on a Bayesian approach to describe car-following behavior of a wide range of vehicles (e.g., human-driven and automated vehicles); and (2) establish a theoretical approximation framework to extend the traditional fundamental diagram to a non-static and stochastic version, based on the calibrated stochastic car-following models. The fundamental knowledge, models, and simulation tool from this research can lead to better design of automated vehicle control strategies and smart traffic management that will ultimately improve traffic throughput, stability, and resilience.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.