Effects of cooperative vehicle dynamics on traffic flow control

Abstract

In the era of advances in vehicular and communication technologies, the need to improve both the efficiency in the use of road capacities and the safety as a consequence of the increasing demand for transport has emerged the realization of traffic management strategies through by Intelligent Transportation Systems (ITS). Having been fed with the methodological advances, the implementation of several intelligent systems, including the Ramp Meters, Variable Speed Limits, and Route Guidance Tools, has therefore become well-known. Furthermore, new advancements in communications, as well as their integration with what recent vehicular technology recommends, increase the usefulness of ITS tools via Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication channels. In this context, the main purpose in a wide range of studies involving traffic flow modeling and control under various configurations has been to determine an ideal rate of penetration for Cooperative Adaptive Cruise Control (CACC) managed vehicles in mixed traffic. According to the related literature, the efficiency in road traffic management can be improved by combining the strategies and hence the tools for ITS in an integrated manner and by improving the control mechanisms used. The emerging research direction in the presence of autonomous road traffic systems contributes significantly to long-term infrastructure planning strategies necessitated by the introduction of connected vehicles into actual traffic. In other words, having projections on market penetration rates of connected vehicles and the effectiveness of varying penetration rates aid in minimizing costly infrastructure investments in the short term by spreading them out over time. Therefore, the main objective in this study has been to improve the performance of the control we employ to mixed traffic flow, which consists of Human Driven Vehicles (HDVs) and CACC equipped vehicles, through strategies for managing freeway traffic. To that end, we theoretically propose a complete control system for freeway networks by combining the RM and VSL tools. Using real data for model calibration, we demonstrate the performance of the controller that we have introduced for the integrated management of freeway traffic flow for the first time in the literature. The modeling and the control feature of this study involves proposing a control framework for freeway traffic, applying this framework to real network traffic simulated using real data, and evaluating the effects of changes in market penetration rate on performance measures such as total system travel time, total throughput, and emission pollutants. We have made use of an H∞ based control strategy to maintain robustness in the presence of disturbances, which could result in poor performance of the system behavior in the presence of exogeneous inputs. One of the primary benefits of such a control strategy is that it can be guaranteed that the dynamical system evolves properly in the absence of disturbances, and that such nominal behavior is preserved up to a steady-state error proportional to the applied disturbance and an H∞ gain in the presence of perturbations. Incorporated H∞ State Feedback Controller aims to sustain the mainstream traffic at a critical density through an integrated management system of freeway traffic. This is achieved by taking into account the queue lengths on on-ramps and the downstream and upstream occupancies on the mainstream in the presence of disturbances caused by inflows to the mainstream and from the on-ramps. In contrast to other robust control systems, H∞ based control is commonly used when working with linear models. We have made use of the Lighthill-Whitham-Richards (LWR) ordinary differential equations (ODE) model to describe traffic flow. The integrated control performance evaluations are carried out on the micro-simulation environments SUMO and PTV VISSIM using the features of a real network and model parameters calibrated based on field data. In order to simulate human drivers, Intelligent Driver Model (IDM) and Wiedemann 99 car-following models are used. For CACC vehicles, Milanes and Shladover's car-following model is utilized. The contribution of the work presented in this thesis is unique in two ways: using an H∞ State Feedback Controller for the integrated control involving RM and VSL; using real data to perform the integrated control for freeway traffic through RM and VSL and, discussing the effects of the combined control with the H∞ State Feedback Controller on the penetration rates of CACC equipped vehicles in mixed traffic.