Graduation Year

2022

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Civil and Environmental Engineering

Major Professor

Xiaopeng Li, Ph.D.

Committee Member

Fred Mannering, Ph.D.

Committee Member

Yu Zhang, Ph.D.

Committee Member

He Zhang, Ph.D.

Committee Member

Anton Kleywegt, Ph.D.

Keywords

Exact Solution, Field Experiments, Heuristic Approach, Optimal Control, Time Geography Theory

Abstract

Emerging connected and autonomous vehicle (CAV) platoon technology holds great potential in improving mobility, enhancing riding comfort, reducing fuel consumption, and mitigating congestion. It reduces car-following gaps via cyber connections. Further, it smooths vehicle trajectories via automated control. Smoothed trajectories and improved aerodynamic efficiency from reduced car-following gaps improve fuel and space efficiency, thus improving traffic performance.

Motivated by such promising advantages, this dissertation investigates trajectory optimization for two basic CAV platoon operations, i.e., platooning and split, which are the premise to implement this technology to benefit society. Two two-stage optimization problems are formulated to solve the platooning and split operations. The first-stage objective guarantees operation efficiency, and the second-stage objective assures riding comfort and fuel efficiency. A feasible cone method is proposed to reveal theoretical properties on the solution feasibility and solve the first-stage problems analytically. Exact and heuristic solutions are proposed to solve the second-level problems, respectively. The feasible cone method is also used to construct valid cuts to expedite the exact solution efficiency in solving the second-stage problems.

Numerical experiments show that heuristic solution approaches can always solve the problem instantaneously without much loss of solution optimality to satisfy real-time application needs. In contrast, the exact solution approach with a state-of-the-art solver may take a much longer solution time that may impose challenges to certain real-time applications. It is also noted that the exact solution time is significantly reduced after accommodating the reduced feasible region. This is critical for applications requiring absolute optimality when the heuristic solution approach fails to reach the exact optimum. Simulation results also demonstrate the superiority of the heuristic approaches in optimizing platooning and split trajectories compared with benchmarks. Sensitivity analysis results shed insights in advising parameter selections of platoon-related logistics to balance the tradeoff between operation efficiency and cost.

Inspired by the simulation results, reduced-scale physical tests are conducted to verify the effectiveness of the proposed heuristic platooning and split approaches. A cascade online learning-based vehicle feedback control structure is proposed to control vehicle movements to follow the designated trajectories solved by the heuristic approaches such that the platooning and split operations can be finished optimally and coordinately. Reduced-scale robot cars are used because they require fewer resources and do not impose any safety concerns. Both longitudinal and lateral movements of the robots are regulated with optimal control techniques. Field experiments demonstrate that vehicles can precisely follow the desired trajectories to complete the intended operations with the proposed control structure. It is noted that the proposed control structure is adaptive because the online learning component can automatically account for the changes of the vehicle running conditions (e.g., roadway and weather). This innovative development greatly saves resources because vehicle motor calibration efforts (i.e., converting the desired speed to the motor speed) are saved by the reinforcement learning agent, which constructs and maintains a dynamic conversion as vehicles operate. This enhances the transferability of the proposed trajectory control structure.

Overall, this dissertation provides theoretical insights in designing vehicle platooning and split trajectories and field evidence in testing the two operations to facilitate future CAV platoon technology implementation and innovation.

Download

Share

COinS