Project Description

Up to now, building data for existing and new buildings has been recorded manually by the client and contractor through time-consuming and laborious inspections. ML-BAU-DOK+ uses information extraction to collect data as automatically as possible in order to reduce the amount of personnel required and at the same time optimize the quality of the recording, taking into account table, text and plan documents.

The aim is to accelerate and improve the FM start-up phase as well as energy efficiency and life cycle analyses. The documents for training the ML algorithms are provided by practice partners. In addition, freely available documents from England are used.

The building data collection includes the TGA systems in accordance with DIN 276, the floor areas in accordance with DIN 277, gif or WoFlV, compliance with operator responsibility regulations and the organization of the documentation in accordance with GEFMA 198 or 924. 

Research Team

• Unit of Real Estate Studies
• Applications of Machine Learning (Prof. Sebastian Vollmer) at RPTU
• Data Science and its Applications (Prof. Vollmer) German Research Center for Artificial Intelligence
  (DFKI), Kaiserslautern
   

Project Partner

• AIF Kapitalverwaltungs-AG, Stuttgart; aif.capital
• alstria office REIT-AG, Hamburg; alstria.de
• Architrave GmbH, Berlin; architrave.de
• IBC Ingenieurbau-Consult GmbH, Mainz; /www.ibc-ing.de
• Stiftung Kloster Eberbach, Eltville; kloster-eberbach.de

Link to Project Website Zukunft Bau. 

Project Term

July 2024 - January 2027