PITA Fiscal Year 2009 Projects - Innovative Infrastructure System
   Assessment Technology

Automated Reverse Engineering of Buildings: Modeling the Building Envelope

Principal Investigators: Daniel Huber, Burcu Akinci

Most work on building information models (BIMs) today focuses on representing the condition of a facility as it was designed, rather than its condition as built or as used. A facility’s as-built condition can differ significantly from the design and can change over time, yet having an up-todate representation of a facility is critical for most applications. Capturing and representing a high-level 3D model of a facility’s as-built condition is currently a time-consuming and laborious manual process, which limits the domains where these as-built BIMs are practical.

This project will develop algorithms and representations to facilitate the automated creation of as-built BIMs. Using 3D point cloud data from laser scanners as input, our research will focus on geometrically modeling and representing core elements related to a building envelope, such as walls, ceilings, and floors. The result will be a method for automatically transforming raw point clouds into a compact 3D information model of a building envelope, consisting of a small number of components, which can be easily manipulated and analyzed. Key challenges that will be addressed include how to detect non-essential “clutter” objects, how to model and represent occluded regions, and how to accurately estimate component boundaries.

The project represents a newly established cross-disciplinary collaboration between researchers at Carnegie Mellon and Quantapoint, Inc., a Pittsburgh-based world leader in laser scanning services. The CMU team has deep expertise in 3D modeling techniques and BIM representation, which will be combined with the extensive field experience of our industrial collaborators at Quantapoint. The project will partially support two graduate students and an undergraduate intern.

The algorithmic and representational advances produced by this project have the potential to radically transform the manner in which as-built BIMs are created and utilized, resulting in new opportunities and application domains within the A/E/C community and beyond.