Understanding existing Structures in building Planning (USP)
The optimization and operation of existing buildings and facilities have become the main focus areas of sustainable urban development, and account for approximately 80% of all ongoing projects. To enable a more effective retrofit process, respectively, documentation (as the first step of the retrofit process) of the existing built environment through efficient process automation in BIM, it is crucial to achieve a comprehensive understanding of building structures and materials, even when concealed behind captured building surfaces. To address this challenge, we have initiated the Georg Nemetschek Institute (GNI) project NERF2BIM, dedicated to automating semantic segmentation and 3D reconstruction, including information integration of building elements in existing buildings. Additionally, we have launched another GNI project, AIDABI, focusing on geometric and structural data acquisition for 3D reconstruction of bridge infrastructure.
To build upon and significantly advance these GNI initiatives, the USP team will unite forces to develop research on three main areas: A. semantics understanding, B. structure/material understanding, and C. relevant information integration in reconstruction models. This approach will amplify the impact of both GNI projects, with a focus on the main novelty – assumptions and information generation on materials and their transfer into the segmented elements in the planning process through BIM.
Furthermore, USP may advance the development of automated retrofit actions by leveraging this understanding of structures and materials.
By leveraging an existing database of 3D-scanned (geo-referenced points and 2D photos) spaces and buildings (e.g. ScanNet++ collection and TUM2TWIN data set, https://tum2t.win/), alongside in-depth knowledge of common building structures and materials, and incorporating segmentation and 3D reconstruction methods - e.g. the novel Gaussian Splatting and even more advanced methods to be developed in the three project years, USP will advance the understanding of building structures and materials. The project is highly novel and interdisciplinary, with significant relevance to the construction and retrofit sector, which encompasses buildings and public infrastructure and contributes to over one-third of global greenhouse gas emissions. Our approach will offer an automated solution for documenting as the first step of retrofit decision-making.