At its Immenstadt site, Bosch is working on “industrializing” AI solutions. The basis for this lies in a scalable IoT architecture and cross-functional implementation teams. The first use cases have already been successfully deployed.

When the screen on the test bench for ABS systems at the Immenstadt plant of Robert Bosch GmbH lights up red, the assembly employees know that the tested component is faulty. What may sound trivial, means an enormous gain in efficiency when it comes to operation practice. This is because start-up effects can lead to errors that falsify the test result. In order to distinguish actual defects from start-up effects, the inspection time must be increased. The result: a deterioration in cycle times across the entire production line.

In order to minimize these time losses, the company today uses a self-learning system at this point that recognizes fault patterns based on collected data and thus differentiates between relevant and non-relevant fault messages. The high success rate is constantly improved by weekly re-training of the algorithms.


Scalable IoT architecture as a key

A flexible, scalable IoT architecture, in which the machines of the line are directly connected, forms the basis. he collected data first flows into the cloud, where they use machine learning algorithms to train the analysis model to recognize relevant test cases. This model is then used at the machine level to execute the process control in real time. Bosch has automated the process of continuous, independent learning on the basis of meta-models to such an extent that the regular updating of the analysis model used at Edge level is fully automated even without human verification.


Know-how-Transfer through tandem deams

In order to transfer this process quickly and efficiently to other areas and use cases in the company, Bosch uses cross-functional teams consisting of a data scientist and a production engineer. These know-how tandems combine data analytics expertise with in-depth process knowledge to help identify and implement potential use cases for on-site machine learning.


Die Industrialisierung AI

This unique combination of scalable IoT architecture, automated machine-learning models, and organizational anchoring helps reduce implementation effort and extend analytics to other areas of the smart factory. In addition to ABS production, the Advanced Analytics solution is already being used in another manufacturing process – more are to follow. The vision of an industrialized AI is thus within reach.

  • Reduction of the implementation time for new AI use cases to a few weeks.
  • Successful implementation of two use cases at the Immenstadt site.
  • Significant increase in productivity.


Winner Category:
Smart Factory
Industry Sector:
Electronic Components for the Automotive Industry
Employees at plant:

Apply now