Quantifying alignment effects in 3D coordinate measurement
Authors: Patrick C. Hammett, Luis M. Garcia-Guzman, Steven W. Geddes, Patrick T. Walsh.
The use of fixtureless, non-contact coordinate measurement has become increasingly prevalent in manufacturing problem solving. Manufacturers now routinely use measurement systems such as white light area scanners, photogrammetry, laser trackers, and portable laser scanners to conduct studies that require measuring upstream supplier parts, tooling, or inprocess subassemblies. For part measurements in these studies, certified fixtures with alignment features such as tooling balls often are not available. Instead, manufacturers rely on ad hoc part-holding fixtures or measure parts without fixtures and perform alignments mathematically. Here, advancements in software are providing operators with numerous alignment options, and users are actively using this functionality. Naturally, these additional capabilities have led to inconsistencies in the alignment method used across measurement studies, often affecting dimensional results. This paper reviews several common alignment or registration methods and provides a metric to assess systematic alignment error. To demonstrate alignment effects, we present a measurement system study of a moderately complex part requiring an over-constrained datum scheme. We first measure the part using a conventional fixture-based method to establish a baseline for static and dynamic repeatability. We then compare these with results from two mathematically-based iterative alignment methods based on fixtureless measurement. Next, we assess the systematic alignment error between the different fixture/alignment alternatives. We show that for the same basic datum scheme provided on engineering drawings, the systematic alignment error is a far more significant issue for problem solving than the repeatability error or equipment accuracy.