 |
 |
 |
|
Abstract and Manuscript Management System - Abstract Detail |
|
| Main Menu | Abstracts | |
| Speaker: | Ulrich Neuschaefer-Rube |
| Title: | Application of multisensor measurements and sensor data fusion in coordinate metrology |
| Topic Group: | Multi-Sensor Metrology and Sensor information fusion |
| E-mail: | ulrich.neuschaefer-rube@ptb.de |
| Co-Authors: | Markus Bartscher, Uwe Hilpert |
| Abstract: |
The increasing demands on the measurement of workpieces in industrial quality control require a continuous improvement of coordinate metrology. There is a need to decrease the measurement uncertainty for complex workpieces as well as to increase the measurement speed and extend the measurable tasks. One answer to this challenge is the apply multisensor measurements which benefit from the advantages of different measurement principles. The data of different sensors is merged to multisensor data to achieve more extended and/or more accurate data. The multisensor measurements can be applied in different ways. Stitching of data sets facilitates measurements with small uncertainties in bigger fields of view. Furthermore, different features at a workpiece (e.g. holes and free form surfaces) can be measured with the most suitable sensor and measurements with small uncertainty can be used to correct data e.g. of another sensor exhibiting systematical errors. In coordinate metrology, tactile probes, optical sensors and, for the last years, industrial computed tomography (CT) systems are used. All of these measurement systems feature different characteristics and the spectrum of measurable tasks is limited by the measurement principle. Tactile probes facilitate coordinate measurements with small uncertainty, but the number of measured points is mostly limited. Optical sensors are fast, measure in a non-contact way, but the geometry and surface characteristics of the workpiece affect the results. CT systems on the other hand offer a complete non-destructive measurement of the workpiece of inner and outer geometries, but the uncertainty is sometimes not sufficient for parts featuring very low tolerances. To overcome these drawbacks the measurement data may be fused to multisensor data. But to merge the data an alignment strategy is needed. This strategy can be e.g. feature based or topology based. We discuss these strategies and the requirements for their application. The influence of the alignment on the results (e.g. the leverage effect concerning points in bigger distance) is shown. As an application example we describe the multisensor measurement of a cast cylinder head segment with optical sensing, tactile probing and CT. For this real production part we perform an analysis of the measurement errors caused by alignment errors and discuss the influence on the measurement uncertainty. |
|
NCSL International © 1996-2008 All Content. July 26, 2008 05:40 PM |