United States Department of Transportation
i
The Application of Unmanned Aerial Systems In Surface Transportation - Volume II-B: Assessment of Roadway Pavement Condition with UAS
-
2019-12-01
-
[PDF-3.52 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report, April 2018 – December 2019
-
Contracting Officer:
-
Corporate Publisher:
-
Abstract:The purpose of this study was to conduct a literature review of existing studies related to pavement condition analysis using Unmanned Aerial Systems (UAS) and conduct a pilot study to evaluate the applicability of using UAS for pavement condition analysis. Each task was undertaken by a different UMass research team. The literature review conducted by the U Mass Dartmouth team suggested that the use of UAS for pavement condition assessment is still in its infancy with little experience and information available in the literature. Currently, no UAS platform can provide pavement condition analysis as currently obtained by traditional automated and manual methods. Photogrammetry for crack detection appears to be the most promising condition assessment technique for potential integration with UAS. The Mass Lowell team reviewed research related to new algorithms for analyzing pavement condition data with a specific focus on deep learning. The results suggested that many deep neural networks models have been developed to detect cracks from images and achieved considerable success, but with limitations. A pilot study was conducted with UAS collected images at the Fitchburg Municipal Airport and analyzed using Crack IT and two deep learning methods. The U-Net model performed very well given a limited training dataset.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
