Bosch is using generative AI in production

In initial projects at two Bosch facilities in Germany, generative AI is creating artificial images to develop and scale AI solutions for optical inspection and to optimize existing AI models.

Bosch aims to reduce the time required for planning and commissioning AI applications from the current 6 to 12 months to just a few weeks. Following a successful pilot, this service for generating synthetic data will be rolled out to all Bosch facilities.

AI IS BEING USED IN PRODUCTION OPERATIONS

Bosch Chairman of the Board of Management Stefan Hartung stated that nearly half of all Bosch facilities are already using AI in their production operations.

Explaining that with the help of generative AI, they are not only improving existing AI solutions but also laying the foundations for the optimal use of this future technology across their global production network, Hartung said:

"There are sound economic reasons for this move. Depending on the size of the facility and the type of production, AI can be used to achieve efficiency gains and cost savings in the 6 to 7-figure range annually per facility. AI has excellent potential for innovation and can make human work even more productive. As a manufacturing company, an established factory outfitter, and an Industry 4.0 trendsetter, Bosch aims to play a pioneering role in the development and implementation of industrial AI."

Bosch pilot facilities are using AI in production planning, monitoring, and control. For example, at the Hildesheim plant, AI-based data analysis helped reduce cycle times by 15 percent during the production ramp-up of new lines. At the Stuttgart-Feuerbach plant, new algorithms reduced part testing processes from 3.5 minutes to 3 minutes.

Software models for generative AI were developed by Bosch Research and are now being implemented in the field by Bosch facilities. While one facility uses an AI method consisting of synthetically generated images to reliably inspect the welds of copper wires in electric motor production, another focuses on the quality assurance of high-pressure pumps.

Bosch Board of Management member and Chief Digital Officer Tanja Rueckert stated that with generative AI, they are taking the next step in the evolution of AI and taking modern production to a new level.

Emphasizing that they are unlocking the potential offered by the Bosch production network and its approximately 230 facilities in their AI solution development efforts, Rueckert noted:

"We are using new technologies. Generative AI helps to harmonize individualization and scaling. This technology offers us the best of both worlds. Bosch has been digitizing and connecting its own facilities and those of its customers for over 10 years. Now we are combining Industry 4.0 with AI. Connected production provides data, and AI evaluates it. The use of AI will make factories more efficient, more productive, and more environmentally friendly. It carries historical significance in AI. Because industrial production is fundamentally changing with AI."

According to the information provided in the statement, the Feuerbach facility had provided manual quality assurance for fuel injection components for years. The nature and complexity of the products, as well as differences in the structure of the production lines, meant that neither rule-based nor AI-supported optical inspection was possible. The new approach is a scalable generative AI that recognizes product variants and error patterns and takes into account different arrangements and sequences in the production process. This is based on a foundation model developed by the Bosch research center and fed with large datasets from the Bosch production network.

AI INSPECTING INDEPENDENTLY

Synthetically generated data is used to improve the model and customize applications in the field. This is expected to enable AI to inspect components independently and only send them to visual inspectors in cases where it is uncertain. At the Hildesheim facility, synthetically generated images were successfully used for training purposes on the first standard systems in electric motor production.

The human eye cannot distinguish synthetically generated images from real images. With the new approach, the facility expects the project duration to be six months shorter compared to traditional methods and the annual productivity increase to be in the 6-figure euro range. Bosch plans to expand its AI approach to other locations.

Many Bosch facilities have one thing in common: they are not only AI pioneers but also consistently focus on Industry 4.0. Errors are detected at an early stage, machine downtime is kept to a minimum, scrap is reduced, and energy is applied in a focused manner. For example, the Bosch research center developed an AI-based system that detects anomalies and malfunctions in the production process and improves product quality.

This software is currently used in approximately 50 Bosch facilities where more than 2,000 production lines are connected. Many Bosch facilities also use AI in the optical inspection of components. For example, more than 20 facilities use optical machine learning, a solution designed by Bosch's special-purpose machinery unit that helps identify difficult-to-define features such as scratches and chips on surfaces and defects in weld seams.