Robust averaging of signals for triangulation sensors

Optical sensors are increasingly being employed for quality assurance tasks. Such tasks require metrologically safe tools, which means that the measuring uncertainty of the employed sensors has to be quantified, compared and minimized. The measuring uncertainties of optical triangulation sensors have been found to increase up to a factor of ten, if typical industrially used surfaces, such as steel, are probed under unfortunate angles. This paper proposes three techniques for reducing errors incurred in measurements of triangulation sensors when the surface of the specimen is not perpendicular to the incident light beam, and when the characteristics of the specimen surface vary. The relative merits of the three techniques are discussed, enabling choices to be made in respect of sensor cost, size of error reduction and speed of operation. The third and most robust technique involves several detectors viewing from different directions in combination with the use of principal component analysis. Principal component analysis is a statistical method to obtain a robust average of all signals of the several detectors.