BIM stands for the use of 3D models throughout the life cycle of a building. All information about the building is described geometrically and semantically with the help of a BIM model. Information on components, component groups, connections, technical systems and the environment is essential. BIM can also be used to optimize the construction of existing buildings and the operation of buildings. For this purpose, 3D building models of the existing structure (as-built models) are required. Currently, as-built models are only available to a limited extent and the introduction of BIM is therefore very delayed. For example, a large number of bridges will have to be repaired in the next few years to maintain the traffic infrastructure. Furthermore, the useful life of individual bridges could be extended through digital maintenance management based on as-built models with high-frequency monitoring using sensor technology. As-built models are also of great importance for real estate management. Maintenance of buildings as well as maintenance and optimization of technical equipment, holistic energy controlling, ensuring accessibility and planning of new buildings, conversions and extensions all benefit from digital information. As-built models of buildings and infrastructures are thus a prerequisite for a wide range of tasks and services in the construction and real estate industry.

Solution approaches

When creating as-built models, aspects such as data security and data sovereignty are very important, since in some cases security-critical information is accessed. Certain information for processing should not first be transferred to external providers, but should be transferred to as-built models as directly as possible at the respective owner using AI processes. In order to easily network existing and new digital services, to process sensitive data securely and to be able to reliably integrate further services and data at a later stage, the GAIA-X reference architecture is used as a basis. The core technical components of the BIMKIT ecosystem can be divided into centralized basic services and decentralized GAIA-X nodes.

AI services (GAIA-X nodes): For the generation of single or multiple objects of an as-built model, AI methods are trained and provided as GAIA-X nodes, using open interfaces. The AI services get authorized access to data used for generating the as-built models via the GAIA-X data infrastructure.

Data Services (GAIA-X Nodes): For the generation of single or multiple objects of a concrete as-built model (e.g., a building), different and partially anonymized data, some of which are available from different providers, are made available via the GAIA-X data infrastructure. For example, for as-built modeling of a bridge, 2D plans of the operator or point clouds created by a specialist planner are used. The user will retain sovereignty over particularly sensitive data and can share data for common use in a controlled manner.

BIM services (GAIA-X nodes): The generated as-built models are integrated into existing BIM systems for further processing. These can be BIM authoring systems that check and individually extend the as-built models on the basis of an initial as-built generation, or systems that already contain as-built models and require selective additions.

Vendor management: The individual GAIA-X nodes are recorded and listed in a central vendor directory. The self-describing functionalities of the offered GAIA-X nodes are used. Inclusion in the provider directory takes place after the governance rules have been checked and certified. The provider directory is thus also a marketplace for future AI apps.

Identity management: A central user management “Single Sign-on” for users is enabled. User identity authorizations can thus be managed uniformly and integrated into other security structures.

Quality management: In order to generate as-built models, various data must be compiled and evaluated with the AI processes. How well certain input data are suited or which AI procedures deliver which results for which questions is to be documented in a central quality management and made available to the users. It should also be possible to process expert knowledge and integrate it into the AI processes. Quality management is an important prerequisite for increasing transparency with regard to the AI processes used and for ensuring broad acceptance.

Certification: Rules are being developed on how providers of data, AI processes and BIM systems can participate in the overall system, e.g. concepts for identifying and testing the self-describing GAIA-X nodes. The GAIA-X nodes must be certified with regard to interfaces, processes and data security. This enables comprehensive transparency through the identification of audited data protection as well as regulatory criteria of the products and services offered.

Data usage control: The guarantee of data sovereignty is essential for acceptance. Access to data must be configurable and traceable in such a way that data does not have to be transferred to external systems.

Plausibility check: Concepts will be developed that allow the results of AI procedures to be checked and made plausible by means of suitable explanatory models.

In summary, the project aims to achieve the following scientific and technical work objectives:

  • Development of AI methods for the evaluation of heterogeneous as-built information, generation of as-built models and provision as services
  • Concepts for combining and evaluating AI methods for detailed, complete and current as-built models
  • Approaches to configure AI procedures for application-specific generation and testing of as-built models
  • Establishment of an open ecosystem based on GAIA-X, taking into account existing standards and existing BIM systems
  • Concepts for decentralized use of existing data sources considering data sovereignty and data security
  • Development of standardized procedures for the integration of further BIM applications and AI processes