Kontron has developed the Kontron VisionAI Framework, an AI-powered optical inspection platform designed to automate visual inspections in downstream manufacturing and assembly processes. It combines artificial intelligence with conventional image processing to inspect products with frequently changing variants.
Many processes following SMT and THT assembly still require manual visual inspection. Kontron says its high-mix manufacturing operations need a flexible inspection approach, while extensive training can make some AI-based systems impractical for frequently changing products. The framework is designed for production from a batch size of one, without a lengthy training phase, and can learn during operation as new products and variants are added.
Cameras capture images of an assembly, and the AI model analyzes them in real time against target specifications. The system can identify missing components, assembly errors and cosmetic defects. Processing runs locally on the Intel-based KBox A-151-PTL-ROBO, independently of cloud systems, and the solution can be integrated into existing manufacturing processes.
Kontron's testing found that lighting and consistent image quality affect inspection results. Reflections, shadows and changing ambient conditions can make features and defects harder to detect, so the company coordinated image acquisition and lighting through repeated tests. Jochen Gimple, VP Production & Operations Kontron Global, said: “The key factor is therefore not the software alone, but the interaction between AI, camera technology, and lighting.”
Kontron will demonstrate the VisionAI Framework at electronica in Munich from November 10 to 13, 2026, at Hall A1, Booth 449.
