AI in Manufacturing – Creating Relationships Between Man and Machine

Project timeline: Jan. 2021 – Dec. 2023      

Funding: € 5.9 M from the EU funding program Horizon 2020 

Partners: 12 partners in seven EU countries

Project coordinator: Technical Research Center Finland VTT

Tasks of FIT: User needs analysis and scenario definition, (semi-) automatic knowledge discovery for AI modeling, provisioning and deployment management, human-AI collaboration and domain knowledge fusion, communication and security framework, fog data retrieval, access and management in real time

In the knowlEdge project, Fraunhofer FIT is working with 12 EU partners under the management of VTT Technical Research Centre of Finland to develop a safe AI system to improve industrial manufacturing processes.

AI technologies are one of the megatrends of digital transformation. They can be used in many different business areas of the production environment – from logistics, to process and product development, to manufacturing. Production systems can now provide vast amounts of data, however, the classic methods of analysis are reaching their limits. For companies, the transition from pure monitoring to predicting conditions is becoming increasingly difficult. Although AI technologies promise business sustainability and product/process quality, the rules and extensiveness of databases are becoming too complex for humans to use. This requires new research approaches and methods to support humans in decision making, in order to better collaborate with AI systems and to merge with (human-AI) knowledge.

knowlEdge plans to break through the barriers to entry for these AI technologies and unleash their full potential to develop a new generation of AI methods, systems and data management infrastructures. The project aims to develop a secure AI system to improve the manufacturing process. To this end, an AI-centric software architecture will be developed to support agile manufacturing scenarios by capturing from heterogeneous sources in manufacturing environments. In addition, tools for semi-automated data preparation (cleansing, processing) will be developed as well. This is achieved by implementing a comprehensive integration methodology that makes it possible to connect all components in a uniform technological information and communication infrastructure. This makes it possible to analyze all data semi-automatically for the purpose of identifying useful patterns and outliers based on domain expertise without the extensive involvement of data scientists. Here, a decentralized knowledge marketplace with an appropriate description of the content will be created, enabling access to knowledge for both data scientists and process experts.

The project solutions will be tested and evaluated in the following three manufacturing sectors:

  • Plastic parts for the automotive industry
  • Supply chain tracking in dairy production
  • AI video analysis of assembly line

Fraunhofer FIT always puts people at the center of innovation and is committed to responsible and sustainable AI. In the project, we contribute our expertise in Data Science and Artificial Intelligence and Human-Centered Engineering & Design. The project is coordinated by VTT Technical Research Center Finland. The EU Commission is funding the project as part of the Horizon 2020 funding program. 

Official project website: