Search

The promising development towards vehicle connectivity and autonomy will impose substantial societal and economic impact not only on the mobility systems but also on the entire cities. This dissertation will, from the perspective of the traffic operators, design effective traffic estimation and control strategies to maximize the benefits of these technologies for the urban traffic systems. We will propose both methodological frameworks and pragmatic guidelines to address this timely and relevant research topic. The main contributions of this dissertation are three-fold. First, this dissertation is among the first to handle the challenging issues in the transition period where vehicles with various technologies coexist in the traffic system (i.e. conventional, connected, and automated vehicles). We propose more accurate queue estimation methods that better exploit the limited information provided by connected vehicles, and develop efficient and robust control strategies to handle the uncertainties due to low penetration rates. Second, this dissertation develops a novel control framework for large-scale urban traffic networks at both the local and network levels. At the local level, we integrate traffic control at signalized intersections with trajectory design of automated vehicles. At the network level, we develop a multi-scale perimeter control strategy that fills two important research gaps: i) how the macroscopic control decisions can be translated into microscopic variables, and ii) how the control objectives at different levels can be synthesized. Third, this dissertation explicitly integrates priority schemes into the proposed control strategies, considering the interactions between different groups of vehicles (i.e. vehicles corresponding to different transportation modes, or with different occupancies, values of time, priority levels, etc.). Part I addresses traffic estimation in a connected vehicle environment, with a particular focus on queue estimation at signalized intersections. The queue estimation results facilitate trajectory reconstruction and thus provide a holistic picture of the urban traffic system. These results also serve as essential inputs to traffic control and management strategies. In this part, we propose a computationally efficient methodology based on a convex optimization to exploit the information provided by connected vehicles. We fill the research gaps in two aspects. First, we relax the widely adopted assumption of uniform demand in a signal cycle. Second, we further reuse the information provided by upstream intersections in order to improve the performance of the algorithm. Simulation results show that the proposed strategy significantly improves the estimation accuracy with a reasonable solution time (0.8s), sufficient for most real-time applications. Results further show that the proposed algorithm is able to handle scenarios with penetration rates as low as 0.1 and is also robust to measurement noises. Part II and Part III investigate advanced traffic control strategies exploiting the information and flexibility provided by connected and automated vehicles at both the local (Part II) and network (Part III) levels, and develop a multi-scale, multi-modal, and multi-technological control framework to maximize the potential of these technologies in urban traffic systems. In each part, we address the challenging issues in the transition period and study how the priority schemes can be integrated into the control strategies to differentiate different vehicle types. In part II, we develop a bi-level optimization based strategy to integrate traffic signal timing and the trajectory planning of automated vehicles at local intersections. We further develop heuristics to switch between different signal control algorithms as the technology evolves. Simulation results show an evident decrease in the total number of stops and delay even when the penetration rate of the connected vehicles is lower than 50%. We further extend the proposed strategy to account for transit signal priority, considering bus stops and bus schedule. Simulation results show that such extension successfully improves passenger mobility over the original strategy. Results also demonstrate that it is valuable to consider bus stops, especially when they are near-side. In part III, we develop a multi-scale perimeter control strategy based on a Model Predictive Control (MPC) algorithm where the network-level decision can be optimally distributed to local-level perimeter intersections to synthesize the competing objectives of both levels. Connected vehicles are assumed to be the only source of information. Simulation results show that the proposed strategy optimizes the performance at both the network and the perimeter intersections, providing much better outputs than the classical controllers. The multi-scale perimeter control strategy is further extended to a stochastic MPC that explicitly handles the uncertainties due to low penetration rates of connected and automated vehicles. It is shown that the total travel is significantly reduced by applying such stochastic MPC. This work is then integrated with priority lanes to prioritize certain groups of vehicles to maximize social welfare. Results show that by introducing the priority scheme, the social welfare can be improved. The proposed estimation and control strategies in this dissertation provide insights on how the emerging technologies can be employed to build a more efficient and flexible urban multimodal urban transportation system. The findings serve as the cornerstone for some promising directions of future research, including the integration with special infrastructure, new mobility systems, heterogeneous data sources, multiple control strategies (e.g. routing), cyber-security, and interactions with other systems (e.g. logistics). This dissertation can be beneficial for traffic managers, local authorities, practitioners, and the automotive industry.