Efficient workflow transforming large 3D point clouds to Building Information Models with user-assisted automatization

Project description

Initial Situation and Problem Definition: On average, a major construction project exceeds its original budget by 80% and takes 20% longer than planned. Digital technologies promise to overcome this unsatisfactory situation in architecture, engineering, and construction (AEC). Indeed, 3D scanners and Building Information Modeling (BIM) have established themselves as flexible, cooperatively creative digital ecosystems in large-scale construction sites and building maintenance.

With the rapid advancement of 3D measurement technology, 3D scanners now provide more accurate and higher-resolution 3D point clouds than ever before. However, the automated transformation of these datasets, which encompass several hundred million 3D points, into the virtual BIM reality remains an unsolved problem. Time-consuming preprocessing steps, the lack of tools for merging 3D scans, various time-consuming, costly, and error-prone manual steps, and the absence of user-friendly interfaces are the biggest challenges.

Objectives and Degree of Innovation: The goals of the proposed project are to research and design a people-centered, efficient workflow for the highly automated transformation of massive, unorganized, raw 3D point clouds into BIMs. With an interdisciplinary team and the active involvement of future users, the LargeClouds2BIM initiative is developing algorithms and data structures for the progressive real-time visualization of massive point clouds, allowing users to work with such data with minimal lead time.

The robust and precise registration of multiple 3D point clouds using similarity transformations makes data capture less dependent on specific 3D scanner technologies. By linking minimal user inputs with powerful optimization methods from geometry processing, the project aims at a novel approach to user-guided reconstruction of BIM objects from 3D point clouds. Researching flexible and active interfaces to open and proprietary BIM ecosystems represents the final innovative sub-goal of the proposed project.

Anticipated Results and Insights: At the system level, the project aims to research and (on a laboratory scale) conceptually demonstrate the proposed innovative workflow for transforming large 3D point clouds into BIMs. At the component level, the interdisciplinary project team is developing theoretical concepts and prototypical implementations of algorithms and data structures for progressive real-time visualization, similarity-based registration, highly automated reconstruction of BIM objects, and user-friendly, active interfaces to open and proprietary BIM ecosystems.

The final assessment of the entire workflow analyzes to what extent the improved automation, flexibility, and accuracy can achieve the expected cost and time savings of approximately 20-30%.