Date of Award
Doctor of Philosophy (PhD)
Dr. Wayne A. Sarasua
Dr. Jennifer H. Ogle
Dr. William J. Davis
Dr. Joseph M. Burgett
Dr. Christopher J. Post
Roadway safety is a multifaceted issue affected by several variables including geometric design features of the roadway, weather conditions, sight distance issues, user behavior, and pavement surface condition. In recent years, transportation agencies have demonstrated a growing interest in utilizing Light Detecting and Ranging (LiDAR) and other remote sensing technologies to enhance data collection productivity, safety, and facilitate the development of strategies to maintain and improve existing roadway infrastructure. Studies have shown that three-dimensional (3D) point clouds acquired using mobile LiDAR systems are highly accurate, dense, and have numerous applications in transportation. Point cloud data applications include extraction of roadway geometry features, asset management, as-built documentation, and maintenance operations. Another source of highly accurate 3D data in the form of point clouds is close-range aerial photogrammetry using unmanned aerial vehicle (UAV) systems. One of the main advantages of these systems over conventional surveying methods is the ability to obtain accurate continuous data in a timely manner. Traditional surveying techniques allow for the collection of road surface data only at specified intervals. Point clouds from LiDAR and imagery-based data can be imported into modeling and design software to create a virtual representation of constructed roadways using 3D models.
From a roadway safety assessment standpoint, mobile LiDAR scanning (MLS) systems and UAV close-range photogrammetry (UAV-CRP) can be used as effective methods to produce accurate digital representations of existing roadways for various safety evaluations. This research used LiDAR data collected by five vendors and UAV imagery data collected by the research team to achieve the following objectives: a) evaluate the accuracy of point clouds from MLS and UAV imagery data for collection roadway cross slopes for system-wide cross slope verification; b) evaluate the accuracy of as-built geometry features extracted from MLS and UAV imagery-based point clouds for estimating design speeds on horizontal and vertical curves of existing roadways; c) Determine whether MLS and UAV imagery-based point clouds can be used to produce accurate road surface models for material volume estimation purposes. Ground truth data collected using manual field survey measurements were used to validate the results of this research.
Cross slope measurements were extracted from ten randomly selected stations along a 4-lane roadway. This resulted in a total of 42 cross slope measurements per data set including measurements from left turn lanes. The roadway is an urban parkway classified as an urban principal arterial located in Anderson, South Carolina. A comparison of measurements from point clouds and measurements from field survey data using t-test statical analysis showed that deviations between field survey data and MLS and UAV imagery-based point clouds were within the acceptable range of ±0.2% specified by SHRP2 and the South Carolina Department of Transportation (SCDOT).
A surface-to-surface method was used to compute and compare material volumes between terrain models from MLS and UAV imagery-based point clouds and a terrain model from field survey data. The field survey data consisted of 424 points collected manually at sixty-nine 100-ft stations over the 1.3-mile study area. The average difference in height for all MLS data was less than 1 inch except for one of the vendors which appeared to be due to a systematic error. The average height difference for the UAV imagery-based data was approximately 1.02 inches. The relatively small errors indicated that these data sets can be used to obtain reliable material volume estimates.
Lastly, MLS and UAV imagery-based point clouds were used to obtain horizontal curve radii and superelevation data to estimate design speeds on horizontal curves. Results from paired t-test statistical analyses using a 95% confidence level showed that geometry data extracted from point clouds can be used to obtain realistic estimates of design speeds on horizontal curves. Similarly, road grade and sight distance were obtained from point clouds for design speed estimation on crest and sag vertical curves. A similar approach using a paired t-test statistical analysis at a 95% confidence level showed that point clouds can be used to obtain reliable design speed information on crest and sag vertical curves. The proposed approach offers advantages over extracting information from design drawings which may provide an inaccurate representation of the as-built roadway.
Cassule, Leonildo Nelson, "Measurement and Evaluation of Roadway Geometry for Safety Analyses and Pavement Material Volume Estimation for Resurfacing and Rehabilitation Using Mobile LiDAR and Imagery-based Point Clouds" (2022). All Dissertations. 3173.
Author ORCID Identifier