There are no accepted standards for determining how many measurements to take during part inspection or where to take them, or for assessing confidence in the evaluation of acceptance based on these measurements. The goal of this work was to develop a standard method for determining the number of measurements, together with the spatial distribution of measurements and the associated risks for false acceptance and false rejection. Two paths have been taken to create a standard method for selecting sampling points. A wavelet-based model has been developed to select measurement points and to determine confidence in the measurement after the points are taken. An adaptive sampling strategy has been studied to determine implementation feasibility on commercial measurement equipment. Results using both real and simulated data are presented for each of the paths.
A mesoscale dimensional artifact based on silicon bulk micromachining fabrication has been developed and manufactured with the intention of evaluating the artifact both on a high precision coordinate measuring machine (CMM) and video-probe based measuring systems. This hybrid artifact has features that can be located by both a touch probe and a video probe system with a k=2 uncertainty of 0.4 {micro}m, more than twice as good as a glass reference artifact. We also present evidence that this uncertainty could be lowered to as little as 50 nm (k=2). While video-probe based systems are commonly used to inspect mesoscale mechanical components, a video-probe system's certified accuracy is generally much worse than its repeatability. To solve this problem, an artifact has been developed which can be calibrated using a commercially available high-accuracy tactile system and then be used to calibrate typical production vision-based measurement systems. This allows for error mapping to a higher degree of accuracy than is possible with a glass reference artifact. Details of the designed features and manufacturing process of the hybrid dimensional artifact are given and a comparison of the designed features to the measured features of the manufactured artifact is presented and discussed. Measurement results from vision and touch probe systems are compared and evaluated to determine the capability of the manufactured artifact to serve as a calibration tool for video-probe systems. An uncertainty analysis for calibration of the artifact using a CMM is presented.
A mesoscale dimensional artifact based on silicon bulk micromachining fabrication has been developed with the intention of evaluating the artifact both on a high precision Coordinate Measuring Machine (CMM), and on a video-probe based measuring system. A high accuracy touch-probe based CMM can achieve accuracies that are as good as the 2-D repeatability of video-probe systems. While video-probe based systems are commonly used to inspect mesoscale mechanical components, a video-probe system's certified accuracy is generally much worse than its repeatability. By using a hybrid artifact where the same features can be extracted by both a touch-probe and a video-probe, the accuracy of video-probe systems can be improved. In order to use the micromachined device as a calibration artifact, it is important to understand the uncertainty present in the touch-probe measurements. An uncertainty analysis is presented to show the potential accuracy of the measurement of these artifacts on a high precision CMM.