Bosch pilots generative AI synthetic images for optical inspection in two German plants

Overview

Bosch is piloting generative AI and foundation models in two German plants, where synthetic images are used to develop and scale AI solutions for optical inspection. At Hildesheim, synthetic images have already trained the first standard systems in electric motor production; Bosch expects the approach to cut AI application planning, launch and ramp-up from six to twelve months to a few weeks.

The challenge

Planning, launching and ramping up AI applications for optical inspection currently takes Bosch six to twelve months. At the Feuerbach plant, fuel-injection components were inspected manually for years because the nature and complexity of the products and differences in the production lines meant neither rule-based nor AI-assisted optical inspection was possible.

The solution

In initial projects in two Bosch plants in Germany, generative AI creates synthetic images in order to develop and scale AI solutions for optical inspection and optimize existing AI models. The approach builds on a foundation model developed by Bosch research and fed by large data sets from the Bosch manufacturing network, refined with synthetically generated data. At Hildesheim, synthetically generated images have already been used to train the first standard systems in electric motor production; at Stuttgart-Feuerbach, quality assurance of high-pressure pumps is to be AI-based in future.

Reported business value

Bosch expects the approach to reduce the time needed for planning, launching and ramping up AI applications from the current six to twelve months to just a few weeks. The Hildesheim plant expects project duration to be six months shorter than with conventional methods, leading to annual productivity increases in the six figure euro range. These are stated expectations from a pilot, not measured results.