University of Minnesota
Longitudinal Planning and Optimization of Connected Vehicles Operation
Abstract
dc:description.abstractThe rapid advancement of connected vehicle (CV) technologies offers significant opportunities to improve intelligent transportation systems through enhanced traffic prediction, vehicle safety, and energy efficiency. This dissertation addresses critical challenges within connected and autonomous vehicle (CAV) systems by introducing advanced methods for real-time traffic prediction that account for lane-changing maneuvers, individualized red-light running warning systems, and eco-driving optimization for electric vehicles (EVs) equipped with multi-speed transmissions. Connected vehicles (CVs) exchange information with other CVs and intelligent infrastructures, enhancing traffic state predictions like speed and density. This aids in developing planning and control algorithms for CVs, including eco driving controls that rely on future traffic conditions. However, predicting traffic states in real-time over the next 10-15 s is challenging, especially with mixed traffic conditions containing both CVs and an unknown number of human-driven vehicles (HVs). Lane-changing maneuvers further complicate these predictions, particularly around eco-driving vehicles, which typically drive at reduced speeds during red traffic signal phases, prompting HVs to overtake, thus increasing lane changes. To address these complexities, this dissertation integrates a macroscopic Payne-Whitham (PW) traffic flow model with a microscopic car-following model that explicitly accounts for the impact of lane changes on the evolution of traffic states. An unscented Kalman filter (UKF) is employed to estimate traffic states ahead of the ego vehicle using real-time data from CVs. Subsequently, the modified PW model forecasts future traffic states, enabling accurate trajectory predictions for the immediate preceding vehicle. Using these predictions, an eco-driving co-optimization problem is formulated specifically for connected and autonomous electric vehicles (CAEVs), considering vehicle dynamics, powertrain efficiency, and battery management comprehensively. The performance of the developed approach is analyzed using scenarios from both the Simulation of Urban MObility (SUMO) and real-world traffic data. Results show a reduction in traffic prediction error of up to 35.95% and an improvement in energy efficiency for the ego vehicle by up to 13.14%. Furthermore, the proposed algorithm is shown to consistently benefit the ego vehicle under different CV penetration rates. Based on the traffic prediction results, an individualized red light running warning system is developed to reduce red light violations and intersection crashes.The proposed warning algorithm aims to overcome the major shortcomings of existing countermeasures for red light violations, where these systems deliver uniform guidance to every driver, prompting some motorists to ignore or misinterpret the alerts and leaving a persistent safety gap. Unlike existing uniform advisory systems, this novel method provides tailored warnings to individual drivers to avoid the broad guidance approach of most existing systems. Recognizing if a driver will run red lights is highly dependent on signal phase and timing, traffic conditions along the road, and individual driver behavior, the proposed warning system contains three parts. First, the traffic prediction algorithm developed in the previous work is used to predict the traffic conditions along the way to a signalized intersection. Second, an optimization based algorithm that computes a driver-specific warning minimizing the risk of a red light violation, given the predicted traffic states and driver reaction model. Third, the resulting advisory-quantifying the required deceleration-is presented on the in-vehicle display and updated continuously as the vehicle approaches the intersection. Both traffic scenarios from numerical simulations generated by SUMO and real-world road tests are used to demonstrate the proposed algorithm's performance under different conditions by comparing with previous work on red light running warning system. The results show that the proposed system provides more effective and accurate warning signals to drivers. In the simulation, the proposed algorithm cuts the ego vehicle's peak deceleration by up to 72.2% relative to an unguided baseline, greatly reducing the risk of red-light violation. Additionally, this research proposes an eco-driving controller specifically designed for electric vehicles equipped with multi-speed transmissions. Formulated as a co-optimization problem, this controller simultaneously optimizes vehicle speed and powertrain operation to maximize energy efficiency. Constraints derived from the traffic prediction algorithm are used to ensure traffic safety and smooth traffic flow in dynamic environments with multiple signalized intersections and mixed traffic. By simplifying a complex nonlinear mixed-integer problem, the proposed controller achieves computational efficiency suitable for real-time applications. To evaluate its performance, traffic scenarios from both numerical simulations generated by SUMO and real-world road tests are employed. The results demonstrate a notable reduction in energy consumption by up to 11.36% over a 18 km drive. Overall, this dissertation demonstrates the substantial potential of integrating connected vehicle technologies with intelligent vehicle control, advancing transportation safety, efficiency, and environmental sustainability.
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- He, Suiyi
Subjects
dc:subject × 6Rights
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Dc Identifier Citation
- He, Suiyi. (2025). Longitudinal Planning and Optimization of Connected Vehicles Operation. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/280273.