In the bidding process of a sheet metal roofing project roofing contractors currently use architectural drawings, a Building Information Model (BIM) file, or aerial photogrammetry to estimate the roofing area. However, a roof structure is never built to the exact drawing dimensions. Hence, after the project is awarded, the as-built dimensions of every roof plane have to be acquired. These dimensions are then used to roll form and cut sheet metal coils into roof panels. There are a number of surveying methods that are currently being used for this purpose including tape measuring, total station surveying, and measuring using aerial imagery. Tape measuring, as the most common practice, requires teams of workers carrying tape measures and climbing all over the target roof. This is one of the most critical safety hazards, particularly on sloped roofs, and contributes to the very high number of occupational injuries and fall deaths which occur in the roofing industry; it is also laborious and not always accurate. On the other hand, total station surveying and aerial imagery either require expensive equipment and/or trained surveyors which impedes their adaption in small projects or involve significant amount of human intervention which translates to higher labor costs.
CEE Assistant Professor Ioannis Brilakis and his Ph.D. student Habib Fathi, with support from the US National Science Foundation and Metalforming Inc., are validating the ability of a novel videogrammetric framework which aims to eliminate the above discussed inefficiencies by acquiring as-built dimensions of roof structures from a mobile set of binocular video cameras. When using this technology, a roofing contractor will simply collect stereo video streams of a target roof once the roof underlayment has been installed. The captured video is then sent to a server. At this server, different forms of visual features are detected and matched between left and right views of the stereo set up and across consecutive sets of frames; this allows chaining the available information in the video and extracting the corresponding 3D information. A 3D wire diagram of the target roof is finally generated which includes the as-built dimensions of each plane. The detailed dimensions are posted on a customer portal in an Extensible Markup Language (XML) format. The contractor can download this data to a USB flash drive and transfer it to a machine that automatically roll forms and cuts sheet metal coil into precise, ready to install panels at the jobsite.
The breakthrough of this framework is enabling a roofing contractor to (1) acquire as-built dimensions of roof planes using a safer, less expensive, faster, and simpler surveying method which provides the required level of accuracy; and (2) upload the acquired dimensions into a computerized roof manufacturing process at the jobsite and digitally fabricate precise sheet metal roof panels ready for the final installation. This automates the entire on-site sheet metal fabrication process.
Dr. Ioannis Brilakis joined CEE in 2009 and established the Construction Information Technology Laboratory (CIT Lab). The CIT lab focuses on creating, synthesizing, modifying and/or adapting the next generation IT methods and systems needed to automate and improve construction, inspection, and rehabilitation methods. Brilakis’ primary research interests lie at the intersection of machine vision and construction engineering. The long term goal of his research team is to create the tools needed to virtualize real world structures and construction processes by automating the opposite of the design-construction process and digitizing the result of construction into a virtual 3D model.
For more information about this initiative, please visit: http://www.cit.gatech.edu/.