AI-powered quality inspection for the production of coated tools and components

Laborious visual inspection and limited automation

In industrial practice, quality assurance for coated tools and components is often performed through manual visual inspections. This process requires the reliable detection of a wide variety of defects, such as geometric deviations, coating defects, or wear. This inspection process is time-consuming, costly, and heavily dependent on the experience of the inspection personnel. Conventional automated image processing systems reach their limits, particularly when dealing with complex geometries, highly reflective surfaces, and varying coating conditions. At the same time, in practice, there are usually insufficient image datasets available for training AI-based inspection systems.

AI-assisted inspection using real and synthetic data

In the “SimVision” project, the Fraunhofer Institutes IST, ITWM, and IGD are collaborating with TU Wien to develop a scalable inspection concept for AI-based visual quality control in production. A central element is a combined database consisting of real-world images and rule-based synthetic training data. This synthetic data is based on physically and mathematically sound models of surface structure, potential defects, component geometry, and optical appearance. This allows for the realistic representation of a wide variety of surface conditions and defect patterns, even when corresponding real-world data is only available to a limited extent. At Fraunhofer IST, an automated inspection system is being developed that will initially be used for defect detection in diamond-coated cutting tools. In addition, the detection of tool wear as well as further application areas for coated components and tools with PVD and CVD coating systems are being addressed.

Flexible and scalable quality inspection of coated components and tools

Users benefit from significantly reduced reliance on manual visual inspections, greater reproducibility of inspection results, and a flexible, scalable solution for automated quality control throughout the process chain for coating tools and components. SimVision thus lays the foundation for efficient, AI-powered visual inspection in industrial manufacturing.