Stefan Hartmann

Research Associate for Digital Transformation at the Fraunhofer Institute for Mechatronic Systems Design IEM in Paderborn


“The production hall of the future will be intelligent, flexible and sustainable, blurring the boundaries between the physical and digital worlds.”

Stefan Hartmann has been working at the Fraunhofer Institute for Mechatronic Design Technology IEM in Paderborn since 2023. Among other things, he coordinates the activities of the KI.NRW flagship project “Datenfabrik.NRW”. The project aims to carry out pioneering work and shape the digital transformation in a real production environment. Based on the most innovative processes and using artificial intelligence, the factory planning, production, logistics and corporate architecture of the participating user companies will be analyzed and transformed in pilot areas. The necessary adaptations of a company are considered in their entirety and, in addition to strategic issues, the effects on the information architecture, process and work organization in the company as well as the competence requirements of the employees are also considered. The results are intended to serve as a blueprint for the digital transformation of manufacturing companies in NRW and beyond. In this interview with KI.NRW, he talks about the goals and challenges of the project as well as his vision of the production of the future.

Stefan Hartmann ist Wissenschaftlicher Mitarbeiter für digitale Transformation beim Fraunhofer-Institut für Entwurfstechnik Mechatronik IEM in Paderborn.
Stefan Hartmann is responsible for the activities of the KI.NRW flagship project “Datenfabrik.NRW” at Fraunhofer IEM © Fraunhofer IEM

Mr. Hartmann, you have many years of experience in production planning and control. How has the manufacturing industry changed in recent years?

Interestingly, in my previous job I was able to accompany the development of a production facility from the very beginning. After the first machines had been built and the commissioning had taken place, questions about process organization, documentation and the support of the machinery by IT systems quickly arose – that was over ten years ago. The ever-new possibilities of digital process support made it possible to work more and more efficiently. More and more data was produced, but this often remained unused.

At the same time, external requirements were constantly increasing: a shortage of skilled workers, sustainability issues and unstable supply chains are just the most well-known. And this is where the major change comes in: the demands on productivity and flexibility have increased enormously over time, so that the manufacturing industry has been looking for answers to this conflict of objectives and now has real levers at its disposal with the help of digital transformation and AI-supported systems.

With this in mind, what is the task of »Datenfabrik.NRW« and how did the idea for this project come about?

The core task of Datenfabrik.NRW is to think, plan and implement the digital transformation holistically using the examples of our beacon factories CLAAS and Schmitz Cargobull in order to make the results, findings and experiences available to other companies for their individual transformation.

The project was launched at the end of 2021 – even then with the subtitle »AI in the production of tomorrow«. This is so exciting because our current understanding of AI, especially in connection with the developments in generative AI, is completely different from what it was back then. If you then look at the project proposal and see the topics addressed at the time, it is very impressive from today’s perspective how quickly and unpredictably the world of AI has changed thanks to tools such as ChatGPT. But what we already knew back then was that the basis for our work was the increasingly available and larger volumes of data, and the major challenge was integrating the developed solutions into existing processes and company architectures.

What developments do you see on the horizon for the project?

Predicting specific developments feels extremely difficult today for the reasons mentioned. We are currently seeing this with the incredible speed of development of generative AI. Nevertheless, we are managing to develop concrete solutions in all transformation areas, i.e. production engineering, manufacturing, logistics and enterprise architecture management, that serve and support employees in these areas. This is also my message at this point: no matter how AI develops, it will only work in the future with the broad acceptance of employees and therefore our AI use cases must always address the problem areas of employees. This is how we create the basis for good, useful and scalable applications that will also help to strengthen the business location in the future.

Regarding artificial intelligence: where is it already being used in factories today and where do you see the greatest potential in the future?

Often more frequently than many people think and know. Classic areas of application include all aspects of demand forecasting and capacity utilization control, for example in production planning and control or logistics. There are also frequent applications in the field of predictive maintenance, as machines nowadays provide a good database. In addition, there is currently a lot of movement around worker assistance systems. I personally also see huge potential there in the context of generative AI, e.g. by providing virtual assistants at production level to answer various questions.

Where do challenges still arise in the development and implementation of AI potential for production?

Challenges often arise when it comes to the question of »how«, and this is very understandable. Not every company has access to large resources of AI experts. And yet I am happy to pass on the message: Have the courage! Talk to your employees to see if there are already ideas for AI potential. Often, data is already available, and you can carry out initial analyses in fairly simple steps and »learn to run«, so to speak. The most important phase in such projects is business understanding, i.e. the phase of describing a business problem and formulating the corresponding objectives. This is the foundation for successful and accepted AI use cases.

How do you work together with the various project partners from research and industry?

In such a large consortium, positive and constructive cooperation is of course the most important thing. On the one hand, there is the combined strength of the Fraunhofer-Institutes, as each institute contributes its own individual focus, so that we have broad thematic coverage. But without business needs, there is no applied research. This means that the ideas that bring business benefits only emerge in collaboration with the partner companies. And this also leads us to broad transfer, which is a core element of the research project: we prepare everything that we develop in Datenfabrik.NRW in such a way that other companies can also benefit from it. All in all, it can be said that we have ideal conditions here in NRW and, together with innovative research institutions and a mix of global players and SMEs, we can develop real strength in global competition.

What does the implementation of use cases from the project look like?

We generally follow the standard procedure model for data mining projects when implementing our use cases: CRISP-DM. We start with business understanding before we dive into the world of data. Building on this, it is then very important to understand what data is available, what it looks like and what initial findings can be derived from it. Only then do we move on to concrete data preparation and modeling before we finally evaluate the results and implement the use case. A key success factor here is taking small steps forward and, if necessary, taking a step back before getting lost in the data jungle.

In a use case for intelligent workforce scheduling, it happened to us once that we were looking for correlations in the data but never gained any promising insights. At some point, we said: »We need to go back to the drawing board and think about the use case in a fundamentally different way.« And now we are making great progress with this use case and are optimistic that we will develop a solution that offers employees real support in their day-to-day business. In general, we always must consider existing process and IT infrastructures in order to avoid creating isolated solutions that nobody can manage. That’s why corporate IT is always on board.

In your vision: What will the production hall of the future look like?

First and foremost, the production hall of the future will be located in Germany as a business location. Because then we will have succeeded in using data-driven solutions and AI support to make our contribution to efficient, robust and sustainable production that can hold its own against international competition. In my vision, I see a highly automated and networked environment in which humans, machines and data can work together seamlessly. In this ideal scenario, data could be analyzed in real time and time-consuming manual data evaluation would no longer be necessary, allowing people to focus on core value creation. Intelligent tools would help to bring about faster and better human decisions, for example to avoid production errors and overproduction and thus ensure sustainable production.

In short: The production hall of the future will be intelligent, flexible and sustainable, blurring the boundaries between the physical and digital worlds.

Stefan Hartmann is a research associate for digital transformation at the Fraunhofer Institute for Mechatronic Systems Design IEM in Paderborn. As project manager, he is responsible for the activities of the KI.NRW flagship project »Datenfabrik.NRW – Artificial Intelligence in the Production of Tomorrow«. Hartmann has many years of practical experience in the fields of production planning and control, supply chain management and digital transformation.