The vision of Industrie 4.0 has triggered a great dynamic towards networked machines and systems (assets), both within individual production sites and between several manufacturing companies. To date, the focus has been on interoperability between individual IT systems. The increasing acceptance of Industrie 4.0 standards can actually significantly reduce the effort required to network assets. However, the actual economic benefits and the value proposition of Industrie 4.0 for production optimization, product improvement and new business models can only be achieved if the data from the assets is used optimally and automatically, including with artificial intelligence (AI) methods.
Data-based AI applications are suitable for solving and optimizing issues that are difficult to solve using traditional methods, e.g. simulation and detailed planning. But how can engineers' expectations of the predictable and quantifiable performance of classic system solutions be reconciled with the astonishing but not very reliable data-based solution finding of AI systems?
To answer this question, an engineering-like, systematic approach is needed to use AI methods as part of a systems engineering process from development to operations. We call this holistic methodology "AI Systems Engineering".
In order to research and define AI Systems Engineering in more detail, Fraunhofer IOSB heads the Competence Center for AI Systems Engineering CC-KING, funded by the Ministry of Economic Affairs, Labor and Housing in Baden-Württemberg. The Process Model for AI Systems Engineering PAISE® developed there is the world's first process model for AI Systems Engineering. It considers the development and operation of AI systems and combines a linear approach for the overall system with agile development and integration of the subsystems, which can be AI systems.
Fraunhofer IOSB uses AI Systems Engineering for projects in the AI Alliance Baden-Württemberg eG and offers software tools, e.g. PAISE® Toolkit, as well as consulting and development services.