«CHANGING AUTOMOTIVE BODY MEASUREMENT SYSTEM PARADIGMS WITH 3D NON-CONTACT MEASUREMENT SYSTEMS A Comparison of CMM vs. CogniTens Optigo 3D Non-Contact ...»
OSAT - Manufacturing Systems Group December 2003
CHANGING AUTOMOTIVE BODY MEASUREMENT SYSTEM PARADIGMS WITH
3D NON-CONTACT MEASUREMENT SYSTEMS
A Comparison of CMM vs. CogniTens Optigo 3D Non-Contact System
UMTRI Technical Report: UMTRI-2003-43
Patrick C. Hammett, Ph.D., Kenneth D. Frescoln and Luis Garcia-Guzman, Ph.D.
Office for the Study of Automotive Transportation (OSAT) University of Michigan Transportation Research Institute (UMTRI) The University of Michigan 2901 Baxter Road Ann Arbor, Michigan 48109-2150 December 5,2003 December, 2003 OSAT - Manufacturing Systems Group Technical Report Documentation Page
1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.
5. Report Date
4. Title and Subtitle Changing Automotive Body Measurement System Paradigms with December 2003 3D Nan-Contact Measurement Systems 6. Performing Organization Code
8. Performing Organization Report No.
7. Authorjs) UMTRI-2003-43 Hammett, P.C., Frescoln, K.D., and Garcia-Guzman, L.
9. Performing Organization Name and Address 10. Work Unit no. (TRAIS) The University of Michigan Transportation Research Institute
11. Contract or Grant No.
2901 Baxter Road NO02690 Ann Arbor, Michigan 48 109-2150 U.S.A.
12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered General Motors - NAO Small Car 1420 Stephenson Hwy. 14. Sponsoring Agency Code P.O. Box 7025 MC: 480-991-320 Troy, MI 48083
15. Supplementary Notes
16. Abstract This report assesses the measurement system capability of the 3D non-contact measurement system Optigo 200, produced by CogniTens Ltd. The Optigo 200 is portable, manually operated, and uses CCD cameras to create 3D past f
4. Measurement System Analysis Case Studies
4.1 Mapping Process Measurement Error and Static Repeatability
4.2 Gage Repeatability and Reproducibility Analysis
4.3 Optigo Vs. CMM: Feature Correlation and Accuracy
Executive Summary The purpose of this report is to assess the measurement system capability of a 3D noncontact measurement system. Three-dimensional non-contact systems generate thousands of measurements or a cloud of points. For complex-shaped parts, point cloud data provides a means to fully characterize a part shape, as opposed to only measuring at discrete point locations. A main driver for 3D non-contact measurement point cloud data is that it provides an enhanced problem solving diagnostic capability for lower total measurement system costs.
The non-contact system used for this study is the Optigo 200 produced by CogniTens, Ltd. This system is a portable, manually operated non-contact measurement system that uses CCD cameras to create 3D part feature representations that may be compared to part design nominal conditions. To evaluate the effectiveness of this system, Optigo 200 measurement readings were compared to measurements of the same panels using Coordinate Measuring Machine (CMM) systems.
This report examines several potential sources of measurement system variation including repeatability (both static and dynamic), reproducibility, and accuracy. Accuracy measurements are based on comparing Optigo 200 data to CMM measurements for identical automotive stamped parts. The parts used in the case study include a rear compartment side rail, a structural reinforcement, and a hood inner panel. Using these panels, various part features are analyzed including surface points, trim edge points, and hole positions.
The results of these studies suggest that the Optigo 200 is capable of meeting typical industry Gage R&R standards, measured by comparing measurement system capability relative either to historical part variation values or to typical tolerance widths. Overall, the results of the case studies indicate that the average Gage R&R for the Optigo system, in terms of 5.150RaR,is 0.20 (or k 0.12 in terms of k 3sigma metric). This is slightly higher than the results involving the same parts using the CMM. The CMM average Gage R&R, in terms of 5.150RkR, 0.13 (or was k 0.08 in terms of k 3sigma). Still, these data suggest that the Optigo 200 is capable of measuring parts with tolerances of approximately h 0.4 or higher. (Note: Typical surface and trim edge tolerances for sheet metal parts are h 0.5 mm or higher.) More importantly, a correlation and mean bias study between the Optigo 200 system and the CMM suggest a very strong correlation with only a small magnitude difference between OSAT - Manufacturing Systems Group December, 2003 systems. In general, the Optigo system measured within 0.1 mm of CMM measurements at common discrete point locations. This finding demonstrates that the Optigo 200 3D non-contact system may replicate the CMM system.
Although the Optigo system is shown capable of meeting typical industry measurement capability standards, some potential challenges exist. For example, the static repeatability of the Optigo system was not as high as that achieved by the CMM. One explanation is that the CMM uses a physical touch probe moving along an ijk vector path normal to the part surface. This appears to allow for better static repeatability than does the 3D non-contact system, which must virtually calculate the normal-to-surface measurements at discrete point locations. In addition, some small differences between systems may exist due to the location at which a feature is measured. In the case of trim edge points, the Optigo system measures along the top surface of the part whereas a CMM may measure in the middle of the blank. Overall, these differences are considered relatively minor.
The overall conclusion of this study is that the Optigo 200 system represents a viable measurement system alternative for automotive body manufacturing in terms of measurement system capability. In addition, the Optigo system offers significant advantages over traditional discrete point measurement systems in terms of diagnostic capability. The system provides the ability to fully measure a panel shape and thus the potential to make better decisions on how best to resolve downstream process dimensional concerns.
December, 2003 OSAT - Manufacturing Systems Group
1. Introduction Part measurement systems used in automotive body applications largely consist of discrete point inspection. Here, a manufacturer takes a stamped or welded assembly part and identifies key features at specific discrete locations defined by three-dimensional coordinates (X,Y, and Z). Manufacturers then measure these discrete locations relative to part design nominal dimensions. These discrete measurements are typically made using either Coordinate Measuring Machines (CMM) or traditional checking fixtures with electronic data collection bushings and measurement probes (see Figure 1). This approach closely links the cost of checking (e.g., time and resources to measure at a specific location) with the number of discrete points identified. As a result, automotive body manufacturers often seek to minimize the number of discrete inspection points to reduce costs.
Figure 1. Sample Checking Fixture with Discrete Point Measurement Location
One concern with trying to minimize the number of discrete measurement locations is that a manufacturer may not fully comprehend potential problem areas. For instance, body manufacturing problems tend to arise in areas not being checked. In addition, some part characteristics such as flatness, parallelism of a mating surface, or trim edge consistency become very difficult to evaluate using discrete points, as each of these characteristics requires multiple OSAT - Manufacturing Systems Group December, 2003 discrete points to estimate. Another concern with discrete point checking systems is that a manufacturer may not be able to pre-define which discrete point locations are critical during the part design and development process. Thus, manufacturers often incur significant costs during the product lifecycle for adding, changing, or deleting discrete point locations.
One strategy that mitigates some of the above concerns is to utilize Coordinate Measuring Machine (CMM) inspection systems. CMMs provide greater flexibility by allowing addindchanginddeleting inspection points through programming, versus having to make hard tooling changes as required by traditional check fixtures with electronic data collection bushings.
CMM inspection, however, is primarily just a more flexible discrete point measurement system.
It has limited capability in evaluating complex part shapes and other characteristics such as consistency of an entire surface, part radii, or parallelism. For these particular features, it requires a significant number of pre-programmed discrete measurement points. For example, to fully comprehend a door assembly, a manufacturer using a CMM might require pre-programming 200-300 discrete point locations. Unfortunately, measuring a large data set with a CMM may take four to eight hours for a single part, and comprehensive information regarding the part shape still may not be fully represented. In addition, the engineering development and metrology resources needed to determine the desired X,Y,Z location and the respective angle of measurement approach (often defined by an ij,k approach vector) for a large CMM data set of a complex part often makes this activity cost prohibitive. Also of concern with CMM systems are the transportation costs associated with moving parts to a special inspection room.
These traditional check fixture and CMM limitations have resulted in the adoption of non-contact discrete point vision systems. These systems, made popular by Perceptron, are widely used for in-line measurement, particularly in automotive body shops to measure large body sides, underbodies, and main body assemblies. In-line vision systems usually incorporate either several fixed position cameras in a work cell or a few cameras mounted to robots. As an example, a full body might require four robots to access all of the discrete point locations.
Historically, these systems have provided tremendous advantages in terms of ease of data collection for discrete point measurements. Still, they have limitations similar to CMMs in terms of their strong cost and time dependency on the number of discrete point locations. As a result, cost considerations often force manufacturers to measure fewer points than desired.
OSAT - Manufacturing Systems Group December, 2003 Recently, the development of 3D non-contact measurement systems that generate point cloud data offers the potential to replace or augment these traditional discrete point checking systems (e.g., traditional hard tooling check fixtures, in-line vision systems, and CMM systems).
These 3D non-contact point cloud systems offer the flexibility to measure the full part shape (or critical part areas) as well as provide measurements at discrete point locations which typically are necessary to satisfy data sampling requirements for process capability analysis studies.
Figure 2 illustrates sample output from a 3D non-contact measurement system for a hood inner panel using the CogniTens' Optigo 200. For this particular example, only certain areas of the part were measured, as indicated by the colored areas. In addition, the colored balls represent discrete point locations, whereas the remaining areas represent clouds of points. These clouds of
points illustrate the conformance of the part surface to the CAD design nominal values. Note:
Dark blue and dark red represent areas with the largest deviations from nominal.
For automotive body applications, CMMs are the typical standard to which to compare new measurement technology systems. For instance, organizations using in-line vision systems routinely calibrate their process measurements based on correlation studies with CMM systems.
In fact, manufacturers commonly incorporate "mean offsets" in their in-line vision systems to correlate the mean dimensions from the in-process inspection studies to their CMM reports. This practice is common because differences in measurement algorithms and the physical tooling locators used to hold parts during measurement can result in inherent mean biases between systems. The need for these "mean offsets" due to inherent fixture biases represents another measurement system challenge. Although these mean offsets are typically related to fixture differences rather than limitations of vision system technology, most companies desire new technology that matches CMM performance in terms of measurement system variation capability and mean consistency (i.e., to reduce the need for mean offsets).
The purpose of this paper is to assess the measurement system capability of a 3D noncontact measurement system developed by CogniTens (the Optigo 200). This paper first includes a brief discussion of 3D non-contact measurement technology and several potential sources of measurement error in automotive body applications. Next, three case studies are used to assess the measurement system capability and make comparisons between the CMM and Optigo 200.
These case studies evaluate measurement system repeatability (both static and dynamic), reproducibility (thus, Gage R&R) and feature correlation (i.e., measurement system accuracy and biases). The various sources of measurement system error also are by body measurement feature such as surface measurements, edge points, and hole positions.
2. 3D Non-Contact Measurement Technology and the CogniTens System Various systems have been developed for 3D non-contact measurement, such as laser scanners/trackers and photogrammetry-based systems. All of these systems involve the use of structured light to generate part measurements. Gershon and Benady [I] and Mitchell  provide overviews of these systems and a more detailed discussion of the various technologies.