ABSTRACT: Understanding the state of a project in space and through time is the foundation of many construction and operation workfows in the built environment. To do so, contractors and owners have tapped into the use of reality capture and BIM-enabled workfows. The exponential growth in the availability of reality and semantic-rich design data, paired with ever more aggressive delivery schedules on recent projects, has called for introduction of solutions that transcend the current capabilities and venture into bleeding-edge AI territory with the goals of (1) automatically identifying design and plan-related risks to improve decision-making, and (2) achieving better control and higher level of predictability through the real-time analysis of production, resources and jobsite conditions. To address these needs, this talk will present the foundational and translational research efforts based on Computer Vision and Machine Learning that leverage this emerging data to automatically create spatio-temporal models that offer geometrical and semantic understanding of physical assets during construction and use phases. The talk will also describe how these semantic-rich reality models can be twinned against BIM and schedule data —in the form of AI-powered digital twins— such that actionable insights based on deviations between actual and expected performance can be generated so that project teams can take corrective actions. The theoretical foundation of these algorithms and systems will be presented during this talk. Lessons learned will be shared on how these solutions have scaled across 1000s of projects, and what their return on investment looks like, at both project and field personnel levels.

BIO: Mani Golparvar is Professor of Civil Engineering, Computer Science, and Technology Entrepreneurship at University of Illinois at Urbana-Champaign. His work in the area of computer vision, machine learning, and BIM has been the recipient of 2018 Walter Huber Research Prize from ASCE, 2017 National Top 20 Under 40 from ENR; 2016 Dan Halpin Award for Scholarship in Construction and 2013 James Croes Medal for innovation in Civil Eng. from ASCE; among other best journal and conference paper awards. Dr. Golparvar is also the CSO and co-founder of Reconstruct, a spin-off Software-as-a-Service company from UIUC. With $28M+ venture capital backing, most recently led by the Nemetschek Group, along with EquipmentShare/Romulus Capital, STO Building Group, ThorntonTomasetti, SHoP Architects, Syska Hennessy, and Asia Pacific Land, Reconstruct Visual Command Center solution creates reality models from photos and 360 videos and integrates them with BIM and project scheduling to visually track progress, analyze productivity, and proactively detect potential delays using predictive analytics; in turn empowering construction companies to keep their projects on schedule and on budget. Reconstruct has been recognized with awards such as the World Economic Forum construction technology innovation award in 2016.