Automated surface inspection
at the JentnerGroup

JentnerGroup specializes in the entire spectrum of electroplating (= electrochemical cutting of metallic coatings) and surface finishing. The medium-sized company was founded in the 1970s and today coats complex workpieces from the medical technology, aerospace and electrical engineering sectors on behalf of customers. The traditional Pforzheim-based company offers a wide range of galvanic coatings, from nickel plating, copper plating and silver plating to rhodium plating. Customers have the highest quality requirements for the products they finish - this applies to functional elements as well as visible surfaces.

For a long time, the majority of optical inspections in the electroplating industry were carried out manually. This activity is demanding, time-consuming and resource-intensive, even for trained specialists, as the smallest material defects that are difficult to identify and repetitive tasks place great demands on human performance. In addition, it was often difficult to evaluate the collected quality metrics automatically. It was usually uneconomical to use the knowledge gained to optimize the electroplating process. For these reasons, C. Jentner GmbH dedicated itself to the search for and introduction of automated, AI-supported inspection systems, which led to success with elunic's AI-supported quality assurance system, AI.SEE™.

To implement the solution concept, a semi-automated, optical inspection system was realized at an inspection table in the factory with the help of the AI-based quality control system AI.SEE™. Part of the visual quality control and defect detection system is also a self-learning AI system (AI.SEE™ Core), which directly evaluates the incoming images, assigns defect classes and controls further processes. Defects and damage, such as elevations or scratches in the surfaces, can thus be detected automatically. The AI system should at least match manual quality detection and ideally surpass it. In the next step, the process is to be extended to other inspection tables and fully automated by commissioning cobots. With the help of artificial neural networks and deep learning, the model will be further trained to detect all defect classes on all materials and products and to be able to display and analyze newly uploaded images directly.

By automating quality control, error-prone processes can be minimized and frequently occurring sources of error, but also the smallest errors, can be quickly and clearly found and assigned, which human inspectors often overlook. The powerful findings increase quality and operational efficiency.