A new class of speed harmonization strategy with adaptive boundaries: a model predictive control approach in a mixed traffic environment

This paper introduces a novel model predictive speed harmonization (SH) control approach featuring a moving SH control section. This system is designed to enhance bottleneck throughput and reduce discretionary lane changes, thereby optimizing traffic flow and safety. The SH control approach is designed for a mixed environment of human-driven vehicles (HVs) and connected and autonomous vehicles (CAVs), which can continuously collect and share individual vehicle trajectory data. The core of this predictive strategy is a hierarchical control structure that begins with macroscopic models and proceeds to microscopic models. The developed model determines the optimal length of the SH control section and the optimal speed of CAVs. Additionally, it incorporates a speed-difference dampening effect to facilitate more harmonized traffic flow. Another distinctive feature of the control strategy is accounting for potential impacts of lane changes by adopting a stochastic car-following model for HVs. The proposed framework is evaluated through numerical simulations for different levels of CAV market penetration. The SH control strategy is found to reduce total travel time (TTT) by decreasing both the number of vehicles queued at the bottleneck and the number of lane-changing maneuvers, while mitigating backward shockwaves. In addition to examining CAV penetration rates, a rigorous sensitivity analysis is conducted to evaluate the impacts of model parameters, latency, and positioning errors on the efficiency of the macro-micro speed harmonization model.