Machine Learning-Based Tool to Predict the Retroreflectivity of Pavement Markings in the U.S.

Pavement marking retroreflectivity is essential for nighttime driving visibility. However, many state departments of transportation (DOTs) in the U.S. lack resources for the continuous monitoring of retroreflectivity on their roads. In 2022, the Federal Highway Administration established a new rule requiring state DOTs to implement a method for maintaining pavement marking retroreflectivity above minimum levels within four years. This research project was aimed at assisting the DOTs complying with this requirement by developing a predictive decision-making tool capable of estimating pavement marking retroreflectivity for up to 3 years using only the initial retroreflectivity and other project-specific parameters. To achieve this, historical data from the National Transportation Product Evaluation Program was analyzed using 6 machine learning algorithms, with the Random Forest model demonstrating the highest accuracy. While the prediction accuracy declined over time, the model outperformed previous studies, achieving an R² of 0.97 and an RMSE of ±31.12 mcd/m²/lux in the first year. To ensure real-world applicability, 6 field experiments were conducted to assess the impact of workmanship and material properties on pavement marking performance. Key findings were that the paint binder type, pigment volume concentration, and film thickness significantly influenced long-term retroreflectivity, while the glass bead drop rate had a minimal impact. Based on these findings, the study recommends incorporating key material properties into retroreflectivity degradation models, implementing data logging for quality assurance, and updating state DOT specifications to optimize pavement marking durability and visibility.

• Record URL:
  • http://apps.trb.org/cmsfeed/TRBNetProjectDisplay.asp?ProjectID=5239

Language

• English

Project

• Status: Completed
• Funding: $100,000.00
• Contract Numbers:

  Project 20-30, IDEA 237

• Sponsor Organizations:

  National Cooperative Highway Research Program

  Transportation Research Board
  500 Fifth Street, NW
  Washington, DC  United States  20001

  American Association of State Highway and Transportation Officials (AASHTO)

  444 North Capitol Street, NW
  Washington, DC  United States  20001

  Federal Highway Administration

  1200 New Jersey Avenue, SE
  Washington, DC  United States  20590
• Project Managers:

  Jawed, Inam

• Performing Organizations:

  Sam Houston State University

  Huntsville, TX  United States 
• Principal Investigators:

  Mousa, Momen R

• Start Date: 20220111
• Expected Completion Date: 20250115
• Actual Completion Date: 20250115

Subject/Index Terms

• TRT Terms: Machine learning; Pavement performance; Retroreflectivity; Road markings
• Subject Areas: Highways; Pavements; Planning and Forecasting;

Filing Info

• Accession Number: 01790501
• Record Type: Research project
• Source Agency: Transportation Research Board
• Contract Numbers: Project 20-30, IDEA 237
• Files: TRB, RIP
• Created Date: Dec 7 2021 2:34PM