Fraunhofer Institute for Production Technology IPTResearch project "GEPARD"
Increasing variant diversity, a shortage of skilled workers and high costs pose considerable challenges for the production of German companies. In the event of variant changes or even minor product adaptations, documents such as assembly plans or CAM data must be regularly recreated or adapted manually. These adjustments are time-consuming, sometimes highly repetitive and often rely on specialized expertise.
Significantly more efficient creation of data and documents through generative AI
The GEPARD project is developing an assistance system based on generative artificial intelligence (GenKI) that (partially) automates the creation and adaptation of production-relevant documents and data during production start-ups, variant changes and other adjustments. The focus is on work-intensive documents such as assembly and work plans as well as CAM data. The assistance system enables more efficient use of skilled workers, secures implicit know-how, improves responsiveness to customer requests and enhances the integration of less experienced employees. As a result, it not only increases productivity, but also the resilience and adaptability of industrial processes.
Assistance System for pragmatic use for companies
As a concrete result of the project, the following outcomes should be achieved:
- Development of an assistance system prototype and integration as an MVP: A prototype of the GenKI-based assistance system is to be designed, developed, tested and integrated and deployed in the form of MVPs (Minimum Viable Products) at application partners and possibly associated partners for two different use cases.
- Concepts and implementation of modules for data and document processing: There will be modules developed that can integrate common company data and manual input into the model system, for example CAM and SAP data records or specifically requested input from technical experts.
- Creation of methods and processes to ensure suitability for transfer: Methods and processes (e.g. in the form of manuals or plans) should be developed and formalized to enable the company-specific adaptation of the assistance system. These should guide the editing and creation of the database, the adaptation of the model system design and the integration (deployment).
Use beyond the “generic” application of GenKI solutions
The domain-specific application of generative AI in industrial production, which goes beyond generic LLM use, is particularly innovative. GEPARD aims to incorporate heterogeneous, unstructured data sources and systematically integrate implicit, domain-specific empirical knowledge into an adaptive assistance system. The project therefore provides new impulses for the user-centric design of AI-based tools and shows how specialized GenKI solutions can profoundly transform production-related processes.
Project consortium
- INC Innovation Center GmbH
- EXAPT Systemtechnik GmbH
- Fraunhofer Institut for Production Technology IPT
Associated partner
- Johnson Electric Aachen GmbH
- Miele & Cie. KG
- MOSOLF Special Vehicles GmbH
Project sponsor
DLR Project Management Agency
Funding
The “GEPARD” research project is funded by the Federal Ministry for Economic Affairs and Energy as part of the “Generative AI for SMEs” technology program.
Funding code: 01MK25009A