Development of Traffic Control Framework for Connected and Automated Vehicles Using Optimal Control Theory
Hong, Seongah, Civil Engineering - School of Engineering and Applied Science, University of Virginia
Park, Byungkyu, Civil & Env Engr, University of Virginia
The state-of-the-art traffic operations strategies adhere to heuristic approach which uses numerical method to find approximate solutions that are close to the true solutions within a certain range. This is because of the complexity of representing traffic dynamics coupled with human drivers’ behaviors. Obviously, such heuristic approach does not guarantee system optimality and is often not implemented in the field due to their computational burden and the need of calibration efforts entailed to the technique of the algorithm. Various traffic control algorithms have been consistently evolved in a way to improve computational efficiency to realize real-time operations.
Although there have been consistent improvements in effectiveness, the research to develop control strategy attaining true optimality is still lacking. Furthermore, it is still awaiting problem to optimize individual trajectory while considers vehicle platoon system.
This dissertation proposes an analytical approach-based control strategies using Pontryagin’s Minimum Principle with an objective of minimization of control efforts (i.e., minimizing acceleration variations). The key merits of the proposed optimal control algorithm are: (i) it guarantees true optimal strategies; (ii) it is computationally less expensive; and (iii) it optimizes not only individual vehicle’s longitudinal dynamics but also guarantees the optimality in terms of the vehicle platoon.
In particular, two problems are addressed in this dissertation: (i) optimal control on the speed of the automated vehicles before they enter a speed reduction zone on a freeway; and (ii) optimal control on the speed of the automated vehicles to follow the preceding vehicle. The control problem is formulated and solved using Hamiltonian analysis to provide an analytical, closed-form solution that can be implemented in real time. The solution yields the optimal acceleration/deceleration of each vehicle under the hard safety constraint of rear-end collision avoidance.
The developed algorithms are implemented and evaluated using the advanced microscopic simulation environment that is built in this research. A set of scenarios is tested to evaluate the performance in various aspects. The factors considered include traffic volumes and the market penetrations of automated vehicles. To evaluate the performance of the proposed algorithm, existing state-of-the-art algorithms that are comparable to the proposed algorithm are modeled and tested under the controlled conditions.
The optimal control algorithm shows significant improvements in mobility, fuel consumption, and traffic flow stabilization compared to those of the base case and the state-of-the-art algorithms under varying market penetrations of automated vehicles. For both of speed harmonization and the traffic flow stabilization control, the optimal control algorithm performs best under 100% market penetration of automated vehicles. The simulation results show that the travel time improves by 4-28% and the fuel consumption improves by 6-21% for different market penetrations of AVs. As for the traffic stabilization algorithm, the acceleration variations improve by up to 18% and the fuel consumption improves by up to 30% compared to the base case of human driven vehicles for different market penetrations of AVs.
The experimental results in this dissertation demonstrate the feasibility of the control algorithm under mixed traffic of automated vehicles and human driven vehicles and provide quantitative assessment in various aspects of mobility, fuel economy and traffic flow stability compared to the existing practices.
PHD (Doctor of Philosophy)
Optimal Control Theory, Connected Automated Vehicles, Speed Harmonization, Traffic Flow Stabilization, VISSIM
Department of Energy (DOE), USAOak Ridge National Laboratory (ORNL), USANational Research Foundation (NRF), South Korea
All rights reserved (no additional license for public reuse)