BBG GROUP
AI Test Line
Test line for AI-based automation in operation
Together with representatives from Helmut Schmidt University/University of the Federal Armed Forces Hamburg (HSU) and Weidmüller, we have commissioned a scientific test line at our company headquarters in Mindelheim.
The fully automated production line serves as a research platform within the framework of the dtec.bw-funded project “EKI – Engineering for AI-based Automation in Production Environments”. dtec.bw is funded by the European Union – NextGenerationEU.
Research platform for the refinement of automotive glass
The production line can be used to refine various types of vehicle glazing with polyurethane (PUR). It has three robots, several cameras, and more than 200 sensors for data collection. Scientists at HSU want to use the facility in 2025 and 2026 to collect extensive data and test various AI algorithms for their practical suitability under real production conditions.
To this end, AI models were trained for specific use cases - for example, for automatic quality testing of individual process steps and for immediate correction in the event of deviations.
Real-time process monitoring
A key component is the comprehensive recording of relevant process parameters during production and their central storage. With the help of optical recordings and AI-supported evaluation methods, production errors are detected in real time and specifically reworked. Another goal is to continuously increase production efficiency based on data-driven optimizations..
The system consists of the following modules
- 180° rotatable rack for feeding in the unprocessed glass panes
- Centering unit
- Primer station
- Flash-off unit
- Mold carrier swystem BFT-P V9 with PUR encapsulation mold
- Trimming and flash removal
- Laser scanning
- 180° rotatable rack for feeding out the finished glass panes
Identify maintenance needs early on
A specially trained AI module uses the recorded data to identify and indicate the need for preventive maintenance on an as-needed basis. This allows targeted maintenance measures to be implemented instead of blanket, time-controlled intervals - which saves resources and increases plant availability..
Faster product changeovers thanks to intelligent control
The production process is divided into individual work steps and stations, each of which has its own decentralized control system. These are networked with a central control system. When changing products, all you need to do is swap the tool: it is automatically recognized by RFID, whereupon all relevant settings and specifications are automatically adjusted on the individual system modules. The system can thus be converted quickly, reliably, and with minimal effort.