AI-Powered Assistance System for Quality Assurance at ZORGE
The flawless production of sophisticated micro-bonded parts requires absolute precision in quality assurance. Until now, for small and medium batch sizes, inspection has been performed during the rework process via a 100% manual and visual inspection under a camera. For large-scale production, however, this method not only reaches its limits in terms of manpower and cost-effectiveness but also often falls short of the required process capabilities. To increase the efficiency of manual inspection, ZORGE Rubber Solutions is breaking new ground. In collaboration with the “Data Engineering and Consulting” master’s program at Albstadt-Sigmaringen University, an image-based assistance system was developed that is deployed directly at the inspection stations.
The machine learning model acts as a digital eye in quality control. In the first step, the existing inspection environment—including the camera system, lighting, and image quality—was thoroughly analysed. Building on this, the students compiled a product-specific image dataset and trained a machine learning AI software model using typical defect patterns. The MLM system automatically detects deviations in the micro-composite parts and immediately highlights the affected, defective area visually in the camera image.
This targeted support enables quality assurance staff to identify and verify product defects much more efficiently and reliably. As a result, the entire inspection process is noticeably accelerated and becomes more objective.
Thanks to the system’s modular architecture, the solution can also be expanded in the future to include additional products as well as further defect and error variants. For ZORGE, this project marks the beginning of its use of artificial intelligence for process engineering tasks.
For further information, please contact Mr Marjan Keber, Head of Quality & Management Systems.