Robust Registration Algorithm for Performing Change Detection of Highway Bridges Using 3-D Laser Scanning Data

United States transportation infrastructure facilities, such as bridges, currently have a grade of C+ according to the ASCE Annual Report Card. Long-term spatial changes of bridges can be important precursors of serious structural accidents. Visual inspection methods rely on the experience of engineers for assessing the spatial changes of bridges, but subjective manual six change inspection introduces several uncertainties due to the lack of detailed spatial data and comprehensive change analysis methods. Three-dimensional (3-D) laser scanning technology enables change analyses of structures by comparing 3-D imageries collected at different times. The challenge of using 3-D laser scanning in bridge change analysis is that existing algorithms for point cloud registration are for aligning data sets collected in a scene where no changes occur. Applying these algorithms for aligning data sets that contain long-term changes of bridges using data collected from different times require engineers to manually select parts of environments that do not change before the algorithm can reliably assess changes. Some studies tried to setup control network in the field to overcome this challenge, but maintaining a control network that would not change between data collection sessions could be time consuming and difficult for outdoor bridge jobsites. This paper presents a robust point cloud data registration algorithm that accurately registers two sets of 3-D laser scanning data sets collected at different years and contains changes. The results indicate that this new 3-D data registration approach can accurately register the 3-D laser scanning data sets collected from different years for effective bridge change analysis.

Language

  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; References; Tables;
  • Pagination: pp 243-255
  • Monograph Title: Eleventh International Bridge and Structures Management Conference
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 01672056
  • Record Type: Publication
  • Files: TRIS, TRB, ATRI
  • Created Date: Jun 14 2018 10:10AM