PräziLoop

To date, quality assurance in the production of high-precision components has been heavily dependent on downstream measurement processes, which are time-consuming and costly. The PräziLoop research project aims to use artificial intelligence (AI) and explainable AI (XAI) to enable continuous quality prediction directly during the grinding process, even for very small quantities.

The project focuses on modern AI methods that enable precise statements to be made about workpiece quality with little (few-shot) or even no (zero-shot) training data. To this end, high-frequency sensor and control data is recorded during machining, analyzed and transferred to a multimodal Grinding Foundation Model. This model serves as a central element for process optimization, real-time control and continuous learning in the sense of an intelligent closed-loop system.

Particular attention is paid to the user-friendliness and explainability of the AI results: By using XAI technologies and Large Language Models (LLMs), the predictions are made transparent, comprehensible and understandable for specialists even without AI expertise. An explanatory chatbot interface is intended to act as a bridge between the technical complexity of the models and the users in production.

The aim of PräziLoop is to make the use of AI in high-precision manufacturing safer, more efficient and more sustainable. The solutions developed should also be transferable to other machining processes via open interfaces and a modular architecture.

Further information on the project can be found here: External link opens in a new window:https://praeziloop.hs-furtwangen.de/