Leveraging Machine Learning and LiDAR for Predictive Maintenance of Roadway Markings

Efficient maintenance of road infrastructure is essential for durability and safety. Longitudinal pavement markings play a key role in traffic management, yet traditional retroreflectivity assessments—via manual inspection or mobile retroreflectometers—are time-consuming and costly. This study introduces a novel methodology using light detection and ranging (LiDAR) and machine learning to evaluate pavement marking retroreflectivity. Leveraging LiDAR intensity data, we apply regression and Extreme Gradient Boosting (XGBoost) models for proactive quality control. The approach includes data collection, noise reduction, and integration with field-measured retroreflectivity to assess markings accurately. Results show a strong monotonic relationship between LiDAR intensity and retroreflectivity. Additionally, Random Forest classification models evaluate compliance with new Federal Highway Administration standards, enabling timely corrective actions. This methodology highlights the benefits of LiDAR-based, data-driven solutions for transportation asset management—offering scalable, cost-effective tools that enhance safety, reduce project timelines, and support sustainable decision-making for construction firms and public agencies.