Inline Wear Detection in High-Speed Progressive Dies Using Photometric Stereo

¹ Institute for Production Engineering and Forming Machines - PTU, Technical University Darmstadt, 64287 Darmstadt, Germany
² Fraunhofer Institute for Computer Graphics Research IGD, 64283 Darmstadt, Germany
³ Technical University Darmstadt, Karolinenplatz 5, 64287 Darmstadt, Germany
⁴ Computer Vision, Imaging and Data Analysis Group, Hochschule Darmstadt, 64295 Darmstadt, Germany

^* Corresponding author: jonas.moske@ptu.tu-darmstadt.de

Abstract

The progressive forming of complex components, including blanking, deep-drawing, and bending, represents an economically efficient sheet metal processing method. Wear control is crucial for ensuring product quality by preventing defects and minimizing material waste. The integration of optical sensors for qualitative wear detection, driven by advancing digitisation and miniaturisation, supplements conventional monitoring techniques. This study introduces a modular, agile camera-based measurement system that records component geometry and allows cause-specific feedback on wear progression. By assigning anomalies to distinct process stages, the system enhances defect diagnosis. The employed photometric stereo analysis relies on multiple images captured under varying illumination angles. Photometric reconstruction enables the calculation of normal maps, facilitating the assessment of surface characteristics through pixel-wise brightness variations. Deviations from predefined standards allow for the precise identification of irregularities. By delivering real-time, non-intrusive insights into the forming process, this approach establishes a foundation for efficient, reliable, and adaptive manufacturing. Its contributions to intelligent forming technologies enable enhanced process control and quality assurance, advancing the state of modern industrial production. Through the fusion of optical monitoring and computational analysis, the proposed methodology represents a significant step towards data-driven, self-optimising manufacturing systems.