AI-integrated production and 'immature' processes

Rapid innovation as a competitive advantage

The high-wage state of Germany and the state of Baden-Württemberg owe much of their prosperity to the ability of their medium-sized industry to quickly and repeatedly bring innovative products to the world market. In doing so, it is crucial to permanently maintain a scientific and technological lead over international competitors and potential imitators.

The world in which we live and work is not stable, but volatile. Agile, versatile production concepts and the rapid empowerment of processes have the potential to sustain value creation in Baden-Württemberg in the long term.

Prof. Dr.-Ing. Jürgen Fleischer, Head of the wbk Institute Production Technology at KIT

Using the approach of AI-integrated production, the Karlsruhe Research Factory aims directly at systematically developing and expanding such advantages in new, challenging manufacturing processes. The starting point is the increasingly frequent observation that engineers and production planners no longer completely specify new production processes as before. Instead, it must be possible to produce high-quality new products while the manufacturing processes themselves are still being developed.

Mastering 'immature' processes

'Immature processes' thus refer to manufacturing processes that are not yet fully understood and mastered, either because they are new, process new materials, or because it is not understood exactly which process parameters are actually responsible for product quality.

In the Karlsruhe research factory, we are investigating which screws in the process need to be 'turned' so that the quality of the products is and remains consistently high. Based on machine learning methods and measurement and control technology, the machines and systems should ultimately adjust their process parameters themselves. They should be able to react automatically if the quality of the products gradually deteriorates or if the ambient conditions change (high humidity, lower temperatures, etc.).

By means of temporarily additional instrumentation with sensors and actuators as well as machine learning, we can get to know processes faster, regulate them effectively, make them productive at an accelerated rate and optimize them, even if we do not yet understand all the influences and correlations at first.

Prof. Dr.-Ing. habil. Jürgen Beyerer, Institute Director Fraunhofer IOSB

The market can therefore be served much earlier with the new products because immature processes can be industrialized more quickly. With the considerably shorter time-to-market, the target markets for new products can be conquered even before the manufacturing processes have been optimized in terms of production costs.

Bringing processes to maturity - in electromobility, lightweight construction and Industry 4.0

But how can the apparent contradiction between immature processes and production based on them be resolved and at the same time lead the process to maturity? The answer lies in the generic basic idea of first massively instrumenting the immature process with sensors, measurement technology and intervention options. The evaluation of the data obtained in this process using, among other things, current machine learning methods and their comprehensible visualization for the expert make it possible to explore the process, understand it more quickly and improve it in a targeted manner.

At the same time, the initial over-instrumentation creates the conditions for being able to impose a desired target behavior on the immature sub-processes in terms of control technology within certain limits, and thus make them controllable, albeit with initially increased effort.

With this methodology, the Karlsruhe research factory focuses on the application fields of electromobility, lightweight construction and Industry 4.0.


Lightweight design concepts enable materials to be used efficiently and help achieve climate and sustainability targets - in both stationary and mobile applications, such as electromobility.

Prof. Dr.-Ing. Frank Henning, Director of the Fraunhofer ICT

Combined competence: The partners of the research factory

Crucial to the success of the methodology outlined above is the cross-disciplinary collaboration of the three research institutes that jointly operate the research factory. The three partners combine the relevant competencies in materials, production and process engineering as well as in automation, sensor and information technology.

The participating institutes and their competences