Author ORCID Identifier

https://orcid.org/0000-0002-7974-5595

Year of Publication

2020

Degree Name

Master of Science in Civil Engineering (MSCE)

Document Type

Master's Thesis

College

Engineering

Department

Civil Engineering

First Advisor

Dr. Gabriel Dadi

Abstract

Personal injuries and property damage due to the failure of snow-plowable pavement markers which detach from pavement surfaces has led to the development of new all-plastic pavement markers which are located entirely below the planar surface of the pavement. The new all-plastic design pushes existing solutions used to avoid striping over highway reflectors into obsolescence since current solutions operate using electromagnets to sense the metal housings of snow-plowable pavement markers. A replacement solution is currently sought by the highway maintenance industry and three different marker detection methods were developed and tested on real-world highways with both new and aging pavement markers to find that optimal solution. With the developed technologies accruing 106,038 observed data points, it is clear that the ideal solution to marker detection and avoidance is the deployment of a machine vision system operating on a deep learning trained model optimized for the detection of differing types of pavement markers on various pavement surfaces. The machine vision system can be further improved in several areas, the most important of which is the optimization model’s processing speeds such that the system could operate at highway speeds while providing real-time analysis of the integrity of installed pavement markers.

Digital Object Identifier (DOI)

https://doi.org/10.13023/etd.2020.362

Marker_Data_Condensed.zip (11 kB)
Pavement Marker Observation Data

Share

COinS