Project Objective & Overview

This project develops a LiDAR-based framework to automatically detect pavement marking locations and assess their conditions, including retroreflectivity. Pavement markings play a vital role in communicating safety-critical information, such as yield and stop control, to road users. When they fade or deteriorate, they can compromise public safety and expose transportation agencies to legal liability.

To address this need, the project collected LiDAR data across approximately 210 miles of bike lanes in the Greater Boston area using vehicle-mounted, backpack/bike-mounted, and high-density airborne platforms. It also developed models for efficient point-cloud segmentation, pavement-marking extraction, and condition evaluation. All outputs were delivered in GIS-ready formats to support maintenance planning, asset management, and complete streets safety assessments.

Key Results & Findings

• Developed an automated LiDAR workflow for pavement marking detection, point-cloud segmentation, and retroreflectivity estimation.
• Utilized mobile, bike-mounted, and high-density airborne LiDAR to capture both fine-grained markings and corridor-wide features.
• Validated LiDAR-derived retroreflectivity against handheld measurements, confirming strong technical accuracy.
• NOAA LiDAR's low point density was insufficient for bike-lane markings. Mobile LiDAR provided the details needed for stencils and symbols but was sometimes hindered by traffic, parked vehicles, and access constraints.
• Identified high-density airborne LiDAR as a scalable solution for regional pavement-marking mapping.
• LiDAR platform selection requires a balance between resolution and coverage. A hybrid airborne–mobile approach may improve cost efficiency and data completeness in future deployments.
• The developed workflow can be expanded to extract additional complete-street assets.