Abstract

A critical step in production manufacturing is the assessment of whether a whether a part has been made within expected tolerances and free of defects. For production additive manufacturing (AM), this step is particularly challenging and costly. AM parts often have complex geometries that can be difficult to measure using conventional tactile probes. Furthermore, many AM processes have accuracy and the presence of defects that are strongly affected by part geometry, motivating new and scalable measurement strategies. This talk describes research on metrology automation to measure and analyze parts made with AM. Our work is conducted in collaboration with a commercial AM factory using production machines and engineering materials, allowing us to study thousands of AM parts and conditions relevant for end use applications. The research collects vision information from AM parts including photographs, high resolution 2D or 3D scans, and X-ray computed tomography (CT). These images are used to train machine learning algorithms that accurately recognize defects, measure part dimensions, and predict part tolerances. This scalable measurement automation allows us to collect large numbers of part measurements far beyond what is possible with conventional methods, enabling new insights about the AM production environment.

William P. King, Ph.D. is Professor and Ralph A. Andersen Endowed Chair in the Grainger College of Engineering, Department of Mechanical Science and Engineering. He holds courtesy appointments in Electrical and Computer Engineering and Materials Science and Engineering. He is the founder of three companies including Fast Radius Inc., which operates one of the world’s largest additive manufacturing factories and was recognized by the World Economic Forum for its technical achievements. He is a Fellow of ASME , AAAS, APS , IEEE, SME , and the National Academy of Inventors.