Optimal control and coordination of connected and automated vehicles in a mixed traffic environment
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
The emergence of connected and automated vehicles (CAVs) introduces a novel dimension to the mobility paradigm that enables efficient communication and real-time computation of control actions to optimize vehicle performance, traffic efficiency, and other associated benefits. While several studies have shown the benefits of CAVs to improve vehicle- and network-level performance by alleviating congestion at specific traffic scenarios, most of these efforts have focused on 100% CAV penetration rate without considering the interaction with human-driven vehicles (HDVs). The consideration of such utopian scenarios may have facilitated the initial development of CAV technologies, but cannot be realized in current transportation conditions. It is expected that CAVs will gradually penetrate the market and interact with HDVs over the next several years. Therefore, for the CAVs to be deployed en masse, technological advancements are needed to be made considering a mixed traffic environment, where CAVs can safely co-exist with the HDVs in the traffic network. In general, a mixed traffic environment poses significant modeling and control challenges due to the stochastic nature of human-driving behavior and lack of communication. Therefore, the question that still remains unanswered is, “how can CAVs safely interact and co-exist with HDVs?” In this dissertation, I address this question through the development of an optimal control and coordination approach. ☐ In the first part of my dissertation, I explore one extreme of the mixed traffic environment spectrum, which is the 100% CAV penetration case, and address the research gaps in the literature related to CAV coordination at traffic corridors consisting of different traffic scenarios. I propose a vehicle dynamics (VD) controller that yields a closed-form analytical solution to the optimal control problem and minimizes transient engine operation and travel time of the CAVs. %Additionally, I show that the VD controller can be combined with a powertrain control architecture to harness additional energy consumption benefits. The effectiveness of the VD controller is validated through a sequential experimentation methodology that shows real-world improvement in fuel economy and traffic throughput. Furthermore, I investigate the problem of trajectory optimization in the presence of system constraints, which is difficult to solve in real time due to its iterative solution methodology according to the standard Hamiltonian analysis. To this end, I develop a condition-based control framework that can identify the activation of system constraints a priori. The proposed framework can explicitly incorporate the state, control, and safety constraints in its formulation, and derive constrained motion primitives for the CAVs with an efficient, real-time implementable algorithm. ☐ In the second part of my dissertation, I address a mixed traffic environment consisting of CAVs and HDVs, and investigate the implication of vehicle- and network-level control of CAVs. First, I consider the problem of deriving safe trajectories for the CAVs in the presence of HDVs with unknown driving behavior, and propose a predictive control approach that considers the future trajectories of the HDVs to ensure collision safety. Then, I investigate the impact of partial penetration of CAVs from a network-level perspective, which indicates that the higher penetration rates of CAVs improve transportation safety and performance metrics, while lower penetration cases result in traffic congestion. ☐ In an effort to address CAV coordination in a mixed traffic environment to alleviate traffic congestion, I propose a novel control paradigm that leverages the concept of vehicle platooning. I transform the problem of mixed traffic coordination into two hierarchical optimal control problems: (i) a platoon formation problem to indirectly control the motions of the HDVs within the network, and once the mixed platoons are formed, (ii) a platoon coordination problem to lead the platoons through a traffic scenario. To address the former problem, I investigate the feasibility of platoon formation by controlling the CAVs, and propose a comprehensive model-agnostic optimal controller which ensures platoon formation without having explicit knowledge of the human driver behavior. Then, I develop a safety-aware, multi-objective receding horizon controller that considers linear, non-linear, and data-driven prediction models, and enables the formation of vehicle platoon satisfying the system constraints that include enhanced safety. The proposed controller is able to form platoons at low penetration rates while it is robust against a wide range of human driving behavior. Subsequently, to coordinate the platoons formed in (i), I develop a robust, single-level, multi-objective optimal platoon coordination framework that accounts for the effect of delayed communication and system constraints. The closed-form analytical solution of the proposed framework can be implemented in real time with the enforcement of lateral and rear-end collision avoidance constraints in the presence of bounded delays. ☐ The research efforts pursued in this dissertation bridge the gap between the two extremes of the mixed traffic environment spectrum, and thus, have a significant impact on the future of mobility. By adopting the computationally efficient and real-time implementable control framework proposed in this dissertation, CAVs can derive optimal motion primitives that can ensure safe and improved mobility in a mixed traffic environment.
Description
Keywords
Autonomous systems, Connected automated vehicles, Decentralized coordination, Intelligent transportation systems, Mixed traffic environment, Optimal control