Robust Laser-Based Optical Measurement in Industrial Harsh Environments.

Abstract

This thesis reports on the design and experimental assessment of reliable dimensional control using laser based optical devices. Modeling and calibration of such systems, as well as the filtering of the delivered data, has been one of the primary motivations. In particular, modern AI-based algorithms such as Deep Learning and Machine Learning have made these processes faster and easier. Two scenarios were chosen to validate experimentally this work. The former consists of an in-line inspection where complex warm forged revolution parts for automotive propulsion systems are measured. The latter computes the flatness and the surface quality of metal sheets produced by a cut to length production line, where the entire process of unwinding, fattening, cutting, and stacking metal sheets takes place.