Our ultimate goal is to provide automation technologies for bridge engineering life cycle data support, starting from bridge drawings and specifications. Based on our patent-pending PDF2BIM technology (see demo here: https://mediaspace.itap.purdue.edu/media/PDF2BIM+Demo/1_3s1scb0t), 3D industry foundation classes (IFC)-based bridge models can be automatically generated from 2D drawings, and semantic information from the 2D drawings are automatically extracted to enrich the bridge IFC model, which in turn will be used for managing the bridge asset (for existing bridges) or guiding construction robot in performing construction tasks (for new bridges). Related to this line of research we have several completed or ongoing federal and state sponsored research projects:

https://www.nsf.gov/awardsearch/showAward?AWD_ID=2231160&HistoricalAwards=false

https://www.nsf.gov/awardsearch/showAward?AWD_ID=1745374&HistoricalAwards=false

https://www.nsf.gov/awardsearch/showAward?AWD_ID=2222838&HistoricalAwards=false

https://www.nsf.gov/awardsearch/showAward?AWD_ID=2121967&HistoricalAwards=false

https://www.nsf.gov/awardsearch/showAward?AWD_ID=1937115&HistoricalAwards=false

https://rip.trb.org/view/1870407

https://rip.trb.org/View/2410439

As part of the technology suite, we need to find/develop/customize good computer vision models that would be able to automatically identify symbols of interest from engineering drawings.