1 Introduction

Dimensional measuring systems have several key subsystems as shown in Figure 1. For each subsystem there exist multiple competing commercial products. These subsystems do not talk to one another without translation unless they all come from the same vendor. This is called the "interoperability problem," and is known to cause large and unnecessary costs to both users and vendors. What is the cost of the lack of interoperability? Currently this cost is known mostly through anecdotes, and the cost is high and persistent. It occurs broadly among users, but large users experience the problem most acutely, because they tend to gather metrology products from full spectrum of vendors. Failure of interoperability mean s that

• Manufacturing information has to be translated, reinterpreted, and reentered many times, sometimes at great cost, without any value-added (often value-subtracted) in the translation

• Users are stifled on best-in-class since it is more difficult to exploit technology changes

• Vendors with bett er or more appropriate technologies can lose out, since users are locked into particular vendors

• Lack of competition typically increases price of equipment, software, and services

• Vendors must continually support multiple interface languages

• Coordinate measuring machine (CMM) operators and programmers must be trained in multiple, redundant interface languages

There are also indirect costs associated with the failure of interoperability, including the fact that there is still no process control use of CMMs on the factory floor and there is little cross-fertilization across vertical sectors.

NIST is working with the dimensional metrology industry to ensure that the subsystems talk to one another, without translation, even if the subsystems come from different vendors. NIST is helping industry develop common interface languages (commonly called specifications). Once everyone agrees on a common specification, NIST designs and develops tests that are used to verify whether implementations of the specification actually conform to the specification. Ultimately the specification is transitioned into a standard.

NIST involvement is essential to the accomplishment of high quality and highly interoperable metrology systems.

Figure 1: The four key interfaces in commercial dimensional metrology systems, Computer-Aided Design (CAD), Dimensional Metrology Equipment (DME), Inspection Process Plan (IPP), and Reporting&Analysis (R&A). The current internationally accepted specifications or standards for these interfaces are also shown in the graphic, namely, STEP APs for CAD, DMIS for IPP, I++ DME for DME, and DML for R&A.

• NIST’s role and perspective are unique, critically important, and therefore necessary for the production of high quality standards and implementations

• Industry works at approximately the same pace as NIST, and since standards development must keep pace with technology advances, if NIST’s level of involvement is below some threshold, the standards will be obsolete when introduced and all our work will be largely in vain

• Interface specification and test development at NIST has a long history and has much commonality across a variety of domains, allowing the metrology interoperability effort to tap into this pool of knowledge and experience.

The metrology interoperability effort worldwide is seeking to create high quality standards, interoperable implementations based on those standards, and a process whereby standards and implementations can adapt to change. NIST is a key player in this worldwide effort, providing technical leadership and demonstrating initiative, commitment, and technical competence in the pursuit of dimensional metrology systems interoperability. Here is a summary of the key results and impact to date of the NIST effort.

• NIST has created conformance tests for two key subsystem interfaces (DME andR&Ain Figure 1) and they have been approved by industry

• Conformance tests for the dimensional machine equipment (DME) interface using the I++1 DME specification have been through several versions and are now in use worldwide [1]. Partly as a result, high quality implementations of I++ DME are now emerging.

• A worldwide distributed testbed for conducting interoperability tests is now in place and is now engaged in active testing 1I++ is a consortium of mostly European automobile manufacturers, including, Audi, Volkswagon, Volvo, DaimlerChrysler, and BMW

• Through the vehicle of cooperative research and development agreements (CRADA), NIST has obtained several vendor products, including six software applications and two coordinate measuring machines (CMMs), at its own testbed in Gaithersburg, Maryland, USA, which is part of the worldwide distributed testbed

• A live demonstration of interoperability using the I++ DME and Dimensional Markup Language (DML) specifications is planned for the International MachineTool Show(IMTS) in September, 2004, showcasing I++DME and DML implementations by most of the key coordinate measuring machine hardware and software vendors worldwide

• NIST input in standards gap and overlap analysis has provided key overall guidance [2]

• NIST voluminous input in specification analysis [3] has reduced ambiguity in the I++ DME specification. Insufficient NIST involvement produces low quality standards output. There is a whole lot of useful work that NIST can do, and much of it can be done in parallel, meaning that any additional support for NIST can always bring further measurable benefit to industry.

2 Background

In May 2000, major corporations came together at a Metrology Summit at NIST expressing their frustrations concerning the proprietary nature of the metrology industry [4]. All shared a similar experience with their inability to implement efficient quality systems necessary to compete in a global economy. Boeing, Ford, and DaimlerChrysler each supplied examples of excessive and needless costs driven by supplier’s proprietary business models. These costs were identical in all industries observed and escalated as information was exchanged through tier one suppliers.

We found the metrology industry to be behind the technology curve. At the core of the problem is a lack of industry standards required to achieve interoperability.

• CAD data could not directly connect to the metrology system resulting in data re-entry to downstream machine software.

• Machine programs written in a proprietary language were not portable, resulting in duplicate programs and measurement errors.

• Unique inspection programs written by independent programmers provided variable results when measuring the same part.

• Each metrology solution contained proprietary machine software increasing training costs and complexity.

It was apparent the industry would not change without a unified voice from major corporations. Each corporation had developed unique requirements and presented these requirements to metrology suppliers. The metrology suppliers were unable to comply with the large number of independent requests.

To resolve this, a broad base of automotive and aerospace companies from North America and Europe organized creating the Metrology Interoperability Project Team (US) and the I++ committee (Europe) to identify gaps or overlaps to existing standards, develop new standards where gaps existed, harmonize the required standards, and facilitate implementation of metrology suppliers.

We are on the brink of success.

• The resulting standards infrastructure we’ve agreed upon should achieve metrology system interoperability.

• Two new standards have been developed with implementation efforts well under way

• Harmonization efforts are succeeding

• The National Metrology Test Bed has been established with hardware and software supporting interoperability testing.

• The first public demonstration with working systems is planned for Sept 2004

3 NIST support for the I++ DME specification

NIST decided to focus first on the DME interface, since it seemed to be industry’s highest priority interface (of the four interfaces in Figure 1). The industry-accepted specification for that interface is the I++ Dimensional Measurement Equipment (DME) interface specification. The I++ DME is a specification for data exchange between dimensional metrology equipment such as coordinate measuring machines and the software systems that program them and analyze their results. The International Association of Coordinate Measuring Machines (IA.CMM), a vendor group that includes Brown and Sharpe, Zeiss, Mitutoyo, and Renishaw, sponsors the I++ DME specification. Major manufacturers such as DaimlerChrysler, Ford, and Boeing, and third party vendors such as Metromec,Wilcox Associates and Tecnomatix all endorse I++ DME.

NIST has developed several versions of a test suite that verifies correct implementation of the I++ DME specification. The great majority of key software and hardware CMM companies worldwide have used the NIST test suite as an essential tool to enable their I++ compliant implementations. Active users of the test suite among vendor companies include Brown and Sharpe, Zeiss, Mitutoyo, Renishaw, Tecnomatix, LK Metrology, and Wilcox Associates. These vendors, as a group, dominate the worldwide CMM market. Active participants in test suite development among manufacturers include Lockheed Martin, Boeing, Ford, Daimler-Chrysler, and General Electric Transportation.

The test suite consists of software test utilities, test programs, and test artifacts. The test suite has led to substantive improvements in both the quality of the specification itself (by identifying interoperability-destroying ambiguities) and the quality of implementations (by enabling compliance). Dozens of comments made by NIST on the specification have been implemented in the specification. The test suite can be used to test implementations on either side of the DME interface, and it verifies the correct handling of both normal and abnormal conditions. It runs on standard PC systems and is available for download on the web [1]. The current version of the full test suite,Version 2.1, includes over 75MB of source code, test files, and documentation. The test suite has also supported several live testing events between NIST and several locations around the world.

4 The Uniqueness of NIST’s Role

More than four years of intimate experience with metrology interoperability by NIST has revealed that

• Industry works on metrology interoperability standards at roughly the same pace NIST is able to work

• If NIST is unable to support a metrology standards effort, it either does not get done at all, or there is some standards disharmony worldwide, or else interoperability is not achieved due to a lack of high quality conformance tests and success metrics. All this is because NIST has capabilities and perspectives that are fundamentally unique in the metrology industry2. NIST’s unique capabilities include:

• Consistent funding for standards work

• Technical expertise in interface standards work

NIST’s unique perspective includes the recognition of:

• The need for unambiguous interface language specifications

• The need for high quality conformance and interoperability tests

• The level of difficulty needed to generate and maintain such tests

NIST has been a critical player for some time now in this worldwide effort and therefore fulfills an essential role in the realization of interoperability in metrology systems. In order to fulfill this unique role, NIST has been serving the metrology industry² in the following ways:

• Leadership support: consulting on standards, conducting informal surveys, organizing and leading meetings, defining project schedules, helping to develop corporate business cases, organizing and executing public demonstrations, giving lectures, facilitating remote interoperability tests, writing white papers and position documents, etc.

• Test suite development: creating the following: conformance test utilities for both sides of each interface, test artifacts, test programs or files, test metrics, and test documents.

• Standards analysis: identifying gaps and overlaps in existing standards and identifying ambiguous sections of specific interface language specifications³.

• Interoperability research: Anticipating future IO standards needs. NIST have developed a Feature-Based Inspection and Control System (FBICS) and an open architecture CMM controller that together contain many of the components found in commercial metrology systems. These two in-house systems have been used to strengthen in-house expertise. NIST have pursued the feasibility of automatic test case generation for test implementations of our interface languages specifications, both for formal and semi-formal languages.

• Testbed development: Entering into cooperative agreements with dimensional metrology software and hardware vendors to set up at NIST a suite of software and hardware from various vendors in order to develop and verify conformance and interoperability tests, set up and test public demonstrations, etc.

5 Risks and Rewards

NIST’s work has been received enthusiastically by the dimensional metrology industry. However, certain factors place the success of this work at substantial risk:

• The rapid pace of inevitable technology changes, and because those changes require changes in standards, means that, once initiated, any standards development effort has to be completed within an unknown time window. Funding affects the pace of this work and therefore risk is increased as much as the funding is reduced (roughly speaking)⁴.

• It is generally true that industry tends to work on the particular standards NIST has chosen to work on. Furthermore, industry tends to work on standards at whatever pace NIST is able to work on those standards. There is much evidence for this, for example, due to lack of funding, NIST has been unable to spend much effort on the CAD interface and it’s arguable that this lack of effort has led to the very little progress industry has made on this interface in the areas of importance to dimensional metrology.

• If specifications are developed that have substantial ambiguities, lack important functionality, or are not accompanied with high quality conformance tests (all part of the NIST role), the implementations of the specification will not have a high level of interoperability. For example, the Dimensional Measuring Interface Standard (DMIS), in many respects an excellent standard, has been under development for more than ten years. However, even though NIST has been entreated for many years to work on this standard, NIST has never had sufficient funding to do test suite development work on this standard. As a result, DMIS has never been accompanied with conformance tests, and many vendors (even up to the present time), it is well known, claim compliance without being compliant. This has been a huge problem for users when they find they find how costly it is to port DMIS programs to the systems of competing vendors. Such programs must be translated with the loss of time, money, and measurement quality, since errors are invariably introduced.

• There are substantial challenges to harmonization of several standards due to competing interests and competing perspectives worldwide. For example, the DMIS standard and the I++ DME specification are incompatible for certain commands. This incompatibility must be resolved if interoperability is to be achieved. NIST is playing an important role of leadership and facilitation to avert this conflict.

Therefore, both the nature of NIST’s role and the speed at which that role is performed are critical to the success of the process. In spite of these risks, many factors contribute to the assured success of this work, as long as NIST has a sufficient level of funding and sufficient staff:

• NIST work in metrology interoperability has had strong and demonstrable worldwide support, so that it can be said that we are riding the crest of a large wave. If we falter now, it will be difficult to regain the momentum we have built and the trust we have gained.

• The technical work we have to do is quite straightforward…it has low risk…we are certain of its success just as we are certain of strong industry support. We simply need the resources to accomplish the task.

• Much of NIST’s work can be done in parallel. The four key interfaces (CAD, DME, IPP, and R&A⁵) in a metrology system, see Figure 1, can all be worked on in parallel and many aspects of the work required on each interface can also be done in parallel.

• The work itself does not require substantial academic sophistication, meaning that a larger pool of NIST staff is qualified to perform the bulk of the work.

• NIST has discovered from experience that vendors give an effort generally equal to the effort put forth by NIST. This is largely due to the facilitating role that NIST plays, namely, organizing and leading meetings, setting agendas, setting project schedules, establishing and coordinating public demonstrations, etc.

• The dimensional metrology industry has committed itself, with NIST encouragement, to develop standards as fully as possible outside the formal standards development bodies in order to expedite development for most of the interfaces we have been working on⁶.

• The work NIST has been called upon to perform is solidly within the NIST mission statement⁷, since our work promotes more accurate and efficient measurement output, the generation of useful interface standards, and the generation of interface standards development technologies

• The work NIST has been called upon to perform is also solidly within the NIST Manufacturing Engineering Lab (MEL) mission statement⁸

6 FutureWork and its Magnitude

As we mentioned above, both CAD and IPP interfaces have been sorely neglected by NIST at substantial cost to industry. Test suites for 1) DMIS semantics testing and 2) DMIS conformance class testing are essential and long overdue. Code incompatibilities, code rewriting, and measurement errors occur because of this lack of effort by NIST, all at substantial cost to industry. This work is expected to take several full-time equivalent (FTE) years to fully accomplish.

A substantial and challenging effort towards harmonization in the STEP APs relating to dimensional metrology is also greatly needed. The incorporation of inspection process information, tolerances, features, and part geometry into standard formats has been an elusive goal for several reasons, many of them not purely technical, not the least of which is to persuade CAD vendors to actually use currently available STEP standards. Several U.S. manufacturers believe that the biggest "bang for the buck" will be found in solving the CAD interface interoperability problem. This work is especially challenging because of the strong political as well as technical dimension. Therefore, it is also expected to take several FTE years to fully accomplish.

The I++ DME specification and the DMIS standard have some outstanding incompatibilities that must be resolved. This requires negotiations and compromises as well as clever technical solutions. The I++ DME work is going to need consistent support from NIST as the specification nears its final version and interoperability testing is just beginning in earnest.

The DML specification also requires yet a substantial amount of work. An English language specification for the XML schema needs to be completed and analyzed for ambiguity. The first version of the NIST DML test suite has yet to be released, but is expected to be released prior to October, 2004. After that, testing with industry must begin as well as the expansion of the test suite to include all the features in the specification, not just the basic feature set that forms the basis for the current test suite version.

Because of NIST’s unique qualifications for these tasks, industry is looking to NIST for its help to accomplish them.

7 Conclusion

NIST has done a great amount of work towards the realization of dimensional metrology interoperability. Our approach seems to be working even though there are major challenges. A great amount of work still needs to be done. This work is an easy win for U.S. manufacturers. U.S. manufacturers win, since they benefit with savings that are purported to be in the millions of U.S. dollars per year per automotive OEM manufacturer.

Many of the challenges in the metrology interoperability effort can be met through increased funding of the NIST support effort. The consistent (if low) level of support for NIST over these last four years has been responsible for the substantive accomplishments by the entire industry as outlined in Section 1. However, this support is only sufficient to perform roughly one fourth of the work NIST is looked to by industry to perform. Furthermore, because of the possibility of parallel work (within each interface) and the substantial benefit (and lowered risk) to industry of accelerated development, substantial increases in NIST support could easily and profitably be spent and would be of assured benefit to industry.

Finally, metrology interoperability hits a "sweet spot" for NIST efforts and therefore is an easy win for NIST and MEL for the following reasons.

• The relatively low risk of the technical work

• The interface standards are not overly complicated

• The dimensional metrology industry is not too large an industry

• Dimensional metrology subsystems vendors are not dominated by very large players (CAD is the single exception)

• The dimensional metrology standards development community is currently led by a substantial and growing group of key OEMs worldwide

References

[1] http://www.isd.mel.nist.com/projects/metrology_interoperability/resources.html.

[2] Thomas Kramer, et al, “Analysis of dimensional metrology standards,” NISTIR 6847, July 2001.

[3] http://www.isd.mel.nist.com/projects/metrology_interoperability/specfeedback/index.html.

[4] http://www.isd.mel.nist.gov/projects/metrology_interoperability/past_meetings.html.

2 This unique role played by NIST has arisen naturally because vendors and users both have a different set of natural capabilities and perspectives with respect to standards work. For example, metrology systems vendors are generally required to define interface standards because they are the most aware of the precise functional requirements on the interface. However, vendors are generally little concerned about interoperability. Metrology systems customers (OEMs and their tier supplies) are very concerned about interoperability and want to see it realized. However, customers are less knowledgeable about functional requirements. Customers are also very busy with making products and have very little time to work on the technical details of standards and to attend the many meetings required to create quality standards output.

3 At some point NIST may additionally be called upon to actually write all or part of a specification.

4 Working on interface specifications one at a time, reduces the negative impact of reduced funding. This has been NIST’s approach for the last four years of fairlylow funding levels.

5 CAD = Computer-Aided Design, DME = Dimensional Metrology Equipment, IPP = Inspection Process Plan, and R&A = Reporting & Analysis

6 STEP AP standards and the DMIS standard are exceptions, because they predated the AIAG MEPT and are not under MEPT control.