Fraunhofer IPA uses AI to optimise product quality and machine parameters in complex, highly variable high-tech production processes. By combining real‑time machine data, optical inspection and individual local AI models, machine settings are autonomously tuned, reducing scrap, setup time and expert dependence. Examples include laser cutting at Trumpf and resistance spot welding at Audi, where our AI-approach enables 100% in‑line quality control, faster ramp‑up for new variants and stable processes despite material fluctuations.

Zielpublikum: Process manager and decision-maker
Voraussetzungen: basic knowledge of AI and computer vision
Level: Basic

Extended Abstract:
Fraunhofer IPA uses AI to optimize product quality and machine parameters in complex, highly variable high-tech manufacturing processes. We develop product-specific optical inspection systems and use them in conjunction with monitored machine parameters, either directly or through retrofitting. Individual local AI models learn how to readjust parameters when product quality is insufficient. The first use case involves Trumpf, a company that develops lasers for cutting metals of varying thicknesses. Incorrect machine parameterization can lead to defects such as burrs or cavities. With the AI-supported optimization developed at Fraunhofer IPA, all relevant process parameters can now be automatically adjusted within a few iterations – without machine downtime, without manual intervention, and with significantly less waste. In a second use case, we showcase our method in resistance spot welding at Audi, where our AI approach enables 100% in‑line quality control, faster ramp‑up for new variants, and robust processes that remain stable despite material fluctuations.