Development of a Robot-mounted 3D Scanner and Multi-view Registration Techniques for Industrial Applications

Abstract

Optical 3D scanning techniques are increasingly replacing the traditional contact-type methods such as CMMs in industrial metrology applications. Using structured light scanners, field of view ranging several mm can be scanned with high resolution in a single view. However, industrial components can span hundreds of mm or even more and often have complex geometry. In such cases, industrial robotic manipulators are used for articulating the 3D scanner to multiple positions. Point clouds obtained from several views have to be aligned into a global reference frame using multi-view registration techniques. In this paper, development of a 3D scanner based on Digital Fringe Projection is presented and its accuracy is evaluated using calibration artifact. Different multi-view registration techniques for a robot-mounted 3D scanner are discussed and their performance is experimentally evaluated.