Heterogeneous Traffic Flow Signal Control and CAV Trajectory Optimization Based on Pre-Signal Lights and Dedicated CAV Lanes

Ji‐Xiang Wang, Haiyang Yu, Siqi Chen, Zechang Ye, Yilong RenSustainability 2023 · DOI

TL;DR

This paper introduces a novel two-layer traffic management system that uses pre-signal lights and dedicated CAV lanes to optimize signal timing and Connected and Automated Vehicle (CAV) trajectories. The system significantly improves intersection capacity, reduces delays, fuel consumption, and emissions in mixed traffic environments.

What Problem Does This Paper Address?

The increasing presence of both human-operated vehicles (HVs) and CAVs on our roads presents a complex challenge for traffic management. This "heterogeneous traffic flow" introduces uncertainties, especially from HV driving behaviors, which can degrade the efficiency and safety of intersections. Existing traffic control systems often struggle to effectively manage this mix, leading to increased delays, fuel waste, and pollution. This paper aims to bridge this gap by proposing a coordinated approach that leverages the capabilities of CAVs and advanced sensing to optimize traffic flow at intersections.

Methods

The core of this research is a two-layer optimization framework:

  • Upper Layer: Signal Phase Optimization: This layer dynamically adjusts traffic signal phase durations in real-time. It uses the Webster optimal cycle formula as a baseline and then refines it based on the actual number and type of vehicles detected in the "control adjustment zone" (an area just before the intersection).
  • Lower Layer: CAV Trajectory Planning: This layer takes the optimized signal timing from the upper layer and plans individual CAV trajectories. This includes optimizing lane-changing strategies and longitudinal acceleration profiles to minimize travel delay, fuel consumption, and lane-changing costs.

A key innovation is the use of pre-signal lights and a "request for information zone" before the intersection. These allow for early detection of HV steering intentions and provide CAVs with advance information about traffic conditions. Dedicated CAV lanes are also implemented to further streamline CAV movement.

To solve the complex optimization problem, the authors employ an improved cuckoo search algorithm combined with a polar dynamics optimization algorithm (EO algorithm) to efficiently find optimal solutions.

Datasets

The paper does not explicitly mention using pre-existing datasets for training. Instead, it relies on a simulated traffic environment created using VISSIM 4.3 for the HV driving behavior model and custom Python scripts for the proposed model. The simulation parameters are detailed, including vehicle arrival rates, traffic demand scenarios (undersaturated to oversaturated), and CAV penetration rates.

Key Insights and Ideas

  • Pre-signal lights are powerful: By gathering HV steering information before they reach the intersection, we can significantly improve traffic predictability and enable better coordination.
  • Dedicated CAV lanes pay off: These lanes not only benefit CAVs but also indirectly help HVs by reducing overall congestion and conflicts.
  • Two-layer optimization is key: Decoupling signal timing and individual vehicle trajectory planning into a hierarchical structure allows for more effective management of complex heterogeneous traffic.
  • CAV penetration matters, but not always 100%: The study found that an 85% CAV penetration rate yielded the maximum improvement in passing capacity, suggesting that some HV lanes might be underutilized at full CAV adoption.
  • Smooth trajectories reduce emissions: Optimized CAV trajectories, avoiding stop-and-go patterns, directly translate to lower fuel consumption and CO2 emissions.

Key Results

The proposed system demonstrated significant improvements across various metrics compared to a fixed-timing signal control strategy:

  • Intersection Capacity: Increased by 8% for all vehicles, 10% for CAVs, and 5.1% for HVs. Maximum enhancement of 15.27% was observed at 85% CAV penetration.
  • Vehicle Delay: Reduced by 13.5% for all vehicles, 18.4% for CAVs, and 9.26% for HVs.
  • Average Vehicle Speed: Increased by 8.54% for all vehicles, 11.6% for CAVs, and 5.9% for HVs.
  • Fuel Economy: Improved by 12.9% for all vehicles, 17.9% for CAVs, and 8.8% for HVs.
  • CO2 Emissions: Reduced by 13.5% for all vehicles, 19.1% for CAVs, and 9.6% for HVs.
  • Lane Changes: Decreased by 16.38% for all vehicles, 22.15% for CAVs, and 10.6% for HVs, indicating smoother traffic flow.

Limitations and Future Work

The authors acknowledge several areas for future research:

  • Improving Signal Timing: The current signal timing is based on the Webster optimal cycle formula, which could be further enhanced.
  • Unobservable/Uncontrollable HVs: The model assumes a degree of cooperation from HVs. Future work could explore scenarios with less predictable HV behavior.
  • Road Network Level Expansion: The current model focuses on isolated intersections. Extending it to a full road network level is a logical next step.
  • Dedicated CAV Lanes: Further investigation into the optimal configuration and impact of dedicated CAV lanes is suggested, particularly concerning the balance between CAV and HV lane utilization.

Summary updated

Abstract

This paper proposes a control system to address the efficiency and pollutant emissions of heterogeneous traffic flow composed of human-operated vehicles (HVs) and connected and automated vehicles (CAVs). Based on the comprehensive collection of information on the flow of heterogeneous traffic, the control system uses a two-layer optimization model for signal duration calculation and CAV trajectory planning. The upper model optimizes the phase duration in real time based on the actual total number and type of vehicles entering the control adjustment zone, while the lower model optimizes CAV lane-changing strategies and vehicle acceleration optimization curves based on the phase duration optimized by the upper model. The target function accounts for reducing fuel usage, carbon emission lane-changing costs, and vehicle travel delays. Based on the Webster optimal cycle formula, an improved cuckoo algorithm with strong search performance is created to solve the model. The numerical data confirmed the benefits of the suggested signal control and CAV trajectory optimization method based on pre-signal lights and dedicated CAV lanes for heterogeneous traffic flow. Intersection capacity was significantly enhanced, CAV average fuel consumption, carbon emission and lane-changing frequency were significantly reduced, and traffic flow speed and delay were significantly improved.
BibTeX
@inproceedings{wang2023heterogeneous,
  author = {Ji‐Xiang Wang and Haiyang Yu and Siqi Chen and Zechang Ye and Yilong Ren},
  title = {{Heterogeneous Traffic Flow Signal Control and CAV Trajectory Optimization Based on Pre-Signal Lights and Dedicated CAV Lanes}},
  booktitle = {Sustainability},
  year = {2023},
  doi = {10.3390/su152115295},
}
Figure