LICENTIATE THESIS 1996 : 19 L

DIVISION OF MANUFACTURING ENGINEERING ISSN 0280-8242

ISNR HLU-TH-L1996/19-L--SE

Off-Line Programming
of Coordinate
Measuring Machines

Eur. Ing.
David Ian Legge

TEKNISKA
HÖGSKOLAN I LULEÅ
LULEÅ UNIVERSITY OF TECHNOLOGY
 Contents

Page
Contents 1
Preface 2
Introduction 3
Summary of the Papers 5
Paper 1 Integration of Design and Inspection Systems 6
- A Research Review
Paper 2 Semi-Automatic Probe Path Planning for 26
Feature Based CMM Programming.
Paper 3 Integration of Design, Inspection and 35
Quality Management Systems
Epilogue 41

1
 Preface
Since 1990 I have learnt to enjoy the landscape of northern Sweden, which is far
removed from the open moors and the tree clad limestone dales of Derbyshire. It is
enough to know that these are only out of sight, and remain for me to return to should
my future lie in that direction.
I have also had the opportunity to learn a little of the Swedish post graduate education
system, the aims of which, less still the methods used to fulfil these, remain, to a large
part, bound in mystery.
Despite this, my knowledge has widened, although not in the direction that I intended.
This licentiate thesis allows me to share with you some of this knowledge concerning the
ideas and practice which lie behind computer aided design and the off-line programming
of co-ordinate measuring machines. It was written to be read.
When I was younger, I used to rock climb. On one occasion whilst hitching out into
Derbyshire, I was given a lift by another climber. During conversation it turned out that
he climbed routes far harder than those which I could or I even aspired to climb; a real
hero. When I suggested that he must have to train really hard to climb at such a level the
hero replied, ''No, not really, it's just that all the people that I climb with climb at that
standard.'' Nuff said.

Eur. Ing. David Ian Legge

Luleå, April 1996

4
 Introduction
Whilst computer controlled (CNC) machine tools can now work virtually unattended 24
hours a day, the fact remains that manufactured components must be inspected to
ascertain whether they lie within those limits set by the designer. The level of inspection
applied varies from the in-process inspection of key dimensions of every component
(often as the basis of in-process correction of tool offsets and wear) through to sample
inspection of components downstream of the manufacturing operation for the purpose
of statistical process control.
Co-ordinate measuring machines
Co-ordinate measuring machines (CMM’s) are a common piece of equipment used in
manufacturing for inspection of components downstream of a manufacturing process.
CMM’s vary in size from ‘desk top’ variants through to large, purpose built units
capable of measuring motor vehicle and similarly sized structures. The basic function of
a CMM is to allow the x, y, z position of a point on the surface of a component to be
established. The CMM consists of a machine tool like structure, with precision
slideways and scales and some form of sensor to determine the point of contact. A
variety of sensor technologies are in common use; touch trigger probes, contact
scanning probes and a variety of non-contact probes.
Programming of Co-ordinate measuring machines
CMM’s are usually computer controlled and, whilst they can be used interactively as a
measurement tool, are usually programmed in order to carry out repetitive inspection on
large batches or samples of parts. Once a programme has been developed, it can be
‘run’ to automatically inspect a component with little or no manual intervention. CMM
programmes can be developed interactively at the CMM, so called ‘teach-in’
programming, or 'off-line' using a simple text editor or some form of programming aid.
In this respect, CMM’s are very similar to robots or machine tools. Graphical off-line
programming (OLP), which is common for CNC machines, was, for many years
uncommon for co-ordinate measuring machines. One of the early research results of
this work was the development of a CMM OLP system based upon the commercial
software available at Luleå University; functionality which is now an accepted part
of many computer aided design (CAD) or simulation packages.
Analysis of Inspection Results
The results of CMM inspection is, in it’s raw form, simply copious data giving details
about each inspection point. This data must be further processed in order to give
meaningful results. The most typical analysis is the generation of a measured ‘feature’
from this raw data by fitting a circle, cylinder, plane etc., through the measured points.
The measured feature geometry may be then compared to the nominal feature definition
and the difference in position, orientation, size etc., compared to the allowable
tolerance(s) associated with the feature. Geometry with a more complex mathematical
definition, so called free form surfaces, present a more complex analysis problem. This
analysis is typically carried out by the CMM, although the raw data may be output for
transfer to a third party system for analysis, reverse engineering etc. The basic data that
is required for meaningful analysis to be carried out is a mathematical description of the
nominal feature, the raw inspection data and the tolerance applicable to the features
being inspected. The first and last of these is available on component drawings, and
usually keyed into the CMM, whilst the second results from the measurement itself.

4
Increasing Automation through Improved Data Transfer
A significant limitation to increasing the level of automation in OLP tools and in the
software for analysing inspection results has been the lack of the necessary 3D geometry
and tolerance data in a computer interpretable form. Whilst a human may ‘read’ a 2D
drawing and interpret the information presented, computers are not able to do this.
Integrated 3D geometry and tolerance (GD&T) models are now common in high end
CAD systems, but this information is not available in downstream systems such as OLP
systems as data transfer standards, which address the transfer of 2D drawings and 3D
geometry, have not kept up with this development. The standard which comes closest
to satisfying this is need is the dimensional measuring interface standard (DMIS) which
is designed primarily for the transfer of inspection programmes and results between
CAD / OLP and CMM’s. DMIS deals with tolerancing, but is limited in its ability to
handle the geometry of the component as a whole; DMIS works with ‘features’ which
are seldom directly accessible in the CAD model. Conversely, the IGES standard,
which is probably the most widely accepted neutral transfer standard for drawings and
3D geometry, has no provision for tolerance data and deals with geometry as a whole,
rather than at the feature level which is necessary for inspection planning and evaluation
of inspection results.
The principle conclusions that was drawn from this work is that the ability to
transfer both tolerance data and its associated three dimensional model data from
CAD via a common neutral format is important for the further evolution of
computer aided inspection systems. The standard for the exchange of product model
data (STEP) would appear the most likely tool for this as it is now widely accepted as
the next generation neutral standard for CAD geometry, and allows new data
transfer requirements to be satisfied by the further developed of the standard.
Definition, Inspection and Analysis of Free Form Surfaces
Related to computer aided inspection there are two principle areas of current interest;
inspection and analysis of free-form surfaces, most commonly through non-contact
inspection technique, and tolerancing of three dimensional models. These areas are
topical as a result of the knock on effect of technical developments related to CAD and
CMM’s and also as natural progressions of ongoing research work related to inspection
and OLP.
Both the above are of interest to users of inspection technology, especially designers and
manufacturers of products with complex surface geometry. Products incorporating this
kind of geometry are often manufactured ‘indirectly’ through the use of moulds and
dies. These products are becoming increasingly technical and there is a need for both
accurate manufacturing processes and a means of assessing and documenting both
manufacturing process (e.g. tool geometry) and resulting product. These points are
addressed more fully in the following papers as well as being the subject of ongoing
research and development.

4
 Summary of the Papers
This licentiate thesis consists of three papers:
Paper 1 Integration of Design and Inspection Systems - A Research Review.
Paper 2 Semi-Automatic Probe Path Planning for Feature Based CMM
Programming.
Paper 3 Integration of Design, Inspection and Quality Management Systems.
Paper 1 reviews a large proportion of the published work concerning the integration of
design and inspection systems; specifically co-ordinate measuring machines.
Given access to a 3D model of component which includes tolerances set against key
component dimensions, highly automated inspection planning is possible. To date this
level of automation has not been found in commercial systems due to the lack of a
tolerancing function in computer aided design systems. (On a 3D solid or surface model
as opposed to simply having tolerances on a 2D drawings.) Limitations within existing
formats for transfer of this toleranced 3D model to off-line programming systems is
another problem which remains to be overcome. This paper has been accepted for
publication in the International Journal of Production Research.
Paper 2 describes how component probing points and movements between these can
be generated based upon a 'feature' based definition of component geometry.
The techniques described have been implemented in part in a prototype off-line
programming system based upon the robotics simulation system GRASP. This
prototype, known as the Inspection Planning Assistant (IPA), allows semi-automatic
off-line programming to be carried out. IPA has been has been tested against a number
of real CMM programming problems from industry. This work remains unpublished.
Paper 3 discusses the new standard for the exchange of product model data (STEP ISO
10303) and specifically a part of STEP (known as an application protocol or AP) for
inspection process planning, AP219. This paper was presented at the 27th International
Symposium on Automotive Technology and Automation in Aachen, Germany.
Work on STEP AP 219 is currently in a state of abeyance, this paper was one of the
actions taken to raise interest, in this case within the automotive industry, in order that
work on the AP could be continued. The relatively limited number of CMM
manufacturers and off-line programming system developers and the wide range of these
systems installed in the automotive industry place made this an ideal group to facilitate
the development and demonstration of this AP. Although carried through to a draft
proposal for funding within the European ESPRIT programme, it proved impossible to
draw together all interested parties at the time. Despite this, the excersise generated
widespread interest and a network of contacts within key organisations within Europe
concerned with STEP and / or inspection technologies. This network will certainly be of
use in the future.

5
 Paper 1

Integration of Design and Inspection Systems

- A Research Review

6
 Paper 1

Integration of Design and Inspection Systems

- A Research Review.
Eur. Eng. David Ian Legge
Department of Manufacturing Engineering,
Luleå University of Technology, Sweden

Abstract
This article reviews some 75 published papers in areas related to the integration of design and
inspection systems. The principle aim of this integration is the automatic or semi-automatic off-
line programming of co-ordinate measuring machines (CMM). The level of automation possible
depends upon the availability of toleranced geometric model in a computer aided design system.
Building upon this base, numerous techniques have been developed for the creation and
validation of inspection process plans for subsequent execution in a CMM.
Keywords
Co-ordinate Measuring Machines, Inspection, Off-Line Programming, CAD

1. Introduction
To take a component model from a computer aided design (CAD) system and automatically
generate all the information required for down stream activities such as machining, assembly and
inspection represents a utopian situation which is many years away from realisation.
The situation facing manufacturing industry today is far from this ideal, with numerous separate
systems linked, at best, through transfer of files in standard formats or via specially written
routines. Individual systems have often limited or no facilities to help the user or to automate the
process. The linking and automation of the activities involved in the design and manufacture of
components hence provides fertile ground for research and development.
This article reviews a significant proportion of the published research work related to the linking
of CAD, automated inspection process planning for co-ordinate measuring machines (CMM’s)
and evaluation of inspection results and feedback to design. The first part of the article provides
an overview of the development and current use of CMM’s in industry. A brief background to
CAD, and the application of tolerances to geometric models, is followed by a review of research
inspection process planning systems. This includes all the key techniques applied in the
automation of .inspection planning leading to valid inspection routines for a CMM. Brief
mention is also made of the return of inspection data to CAD.
Many of the techniques discussed are now commonly available in commercial systems.
Development and application of is limited by industry demand for off-line programming
techniques, the availability of CAD systems allowing tolerancing of solid geometric models and
by the availability of neutral interface standards for the transfer of these toleranced models to off-
line programming systems. The current pace of development of the standard for the exchange of
product data (STEP), which will allow this data transfer to take place, is likely to lead to increased
automation of many process planning activities, not least of which will be inspection process
planning. Possible future trends are hence briefly discussed at the end of the article.
This article provides a thorough and timely review of the state of the art for those working, or
wanting to work, with off-line programming for co-ordinate measuring machines.

7
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

2. The Use of Co-ordinate Measuring Machines

The purpose of a manufacturing system is to produce parts of specific dimensions and form. The
growing importance of product quality has lead to an increasingly important role being played by
inspection within manufacturing systems.
An inspection system should be able to measure ''the dimensional characteristics of randomly
presented parts of virtually any configuration or complexity and provide real-time feedback to the
manufacturing process.'' [1] Current 'State of the Art' co-ordinate measuring machines (CMM’s)
approach this ideal, being highly automated and often including automatic probe changers,
articulated probe heads, protective enclosure, palletised part handling, automatic part
identification, user friendly control software, and automatic statistical analysis [2] with data
management software also becoming more common. [3]
CMM’s were originally developed as a flexible means of performing fast and accurate
dimensional inspection of prismatic parts. They have since become widely adopted as a general
purpose tool with the capability to measure many different types of component. [4] A CMM
may be used to inspect incoming finished and semi-finished parts as well as for the inspection of
components before, during and after actual production processes. In-process inspection can be a
part of the production equipment but is perhaps more commonly an accompanying activity, for
example in a manufacturing cell. [5] Whilst CMM’s are best located in an environmentally
controlled room, improvements in CMM design, construction materials and isolation techniques
means that they are now commonly placed on the shop floor. This increasing use of CMM’s at
or near individual production activities is reflected in the rapid growth for 'shop hardened'
CMM’s. [6]
CMM’s are often used to indirectly check the capability of production processes and equipment.
[7] Small batch sizes, just-in-time (JIT) manufacturing and highly productive numerically
controlled (NC) machine tools used today requires near instant component verification.
Techniques such as statistical process control (SPC) are not applicable with small batch sizes and
one-off production. In these situations, the use of CMM’s to inspect individual components can
reduce scrap or rework. [8] This is also true even if inspection is for process refinement, for
example when fine tuning the form of moulds and dies. [9]
The most obvious benefits from using CMM’s are the significantly lower inspection times when
compared with traditional inspection techniques; reductions of 90% being commonly quoted. [8]
CMM’s often eliminate the need and cost of complex gauging fixtures, with only simple fixtures
being required to support the component correctly with respect to datum surfaces. In many
cases even this type of fixturing is unnecessary, the component simply being located on the table
of the CMM and datum faces being located as part of the inspection process.
The use of CMM’s is changing the nature of inspection, many different inspection procedures
which were once carried out using different techniques can now be carried out on the CMM. [6]
However, in practice inspection remains a hybrid task with a mixture of traditional measurement
techniques used in combination with a CMM.

3. Geometry Specification for Manufacturing and Inspection

3.1. Computer Aided Design Representations for Nominal Geometry.
3.1.1.Wireframes.
The earliest computer aided design (CAD) systems modelled geometry in the form of points,
lines and arcs in two dimensional space; effectively acting as an electronic drawing board. These
systems naturally evolved into three dimensional geometry. A 3D object modelled in this way is
known as a 'wireframe.' When displayed, a wireframe model is 'transparent' as only edges and
8
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

not the surfaces between these are explicitly represented. However, hidden line removal
algorithms can be applied to give the impression of a solid. [10]
3.1.2. Surface and Solid Models.
One of the representations which explicitly defines the surfaces of a model, rather than just the
framework or outline, is the boundary representation or B-rep. In a B-rep model, the object is
described by the set of faces which bound the object, individual faces are bounded by a loop of
edges which are in turn specified by their vertices; not unlike a wireframe. Neither the boundary
of a face nor the face itself need by linear/planar, more complex surfaces or boundaries can be
described by mathematical curves which are known as splines, the surfaces generated being
known as surface patches. [11] B-rep modellers are generally acknowledged as offering a good
basis for the visualisation of a component at a graphics terminal as the geometry of each surface
is explicitly known.
Another common representation, known as constructive solid geometry (CSG), store component
geometry as a set of Boolean operations, cut, join, intersect etc., applied to a limited set of solid
primitive objects, cubes, cylinders, cones etc. CSG models are a good basis for calculating mass
related properties. However, this data structure, known as a CSG tree, contains information
about an object in an 'unevaluated' form which needs to be evaluated into the explicit vertices,
edges and surfaces of the resulting solid for the purpose of visualisation. [12] In order to benefit
from the advantages of both representations, many of today's CAD systems use a hybrid B-
rep/CSG representation. [10]
3.1.3. Features, Product Data Models and Meta Models.
The basic geometric entities of a component model in a CAD system, such as a surface, edge or
vertex, or an auxiliary geometric attribute of a part such as a centre line, are often referred to as
primitive features. Within the component geometry, specific areas of geometry can be identified
which perform particular functions. These are often known as form features. Examples of form
features would included slots, holes, chamfers etc. [13] Some CAD systems allow geometry to
be created using user defined form features or allow primitive features or surfaces to be identified
and to have additional parameters associated with them for use in down stream activities such as
process planning or NC part programming. These systems are often known as feature based
systems. Techniques are also available for identifying features within none feature-based CAD
systems. [14]
More comprehensive data structures which take account of the overall needs of engineering, not
just nominal and toleranced geometry, are known as product data models, [15] or metamodels.
[16]
3.2. Geometric Tolerancing
3.2.1. Geometrical Tolerancing Standards.
The current ANSI / ISO standard for dimensioning and tolerancing divides geometric tolerances
into four categories; form, orientation, location and runout. Tolerances should only be applied
''where they are essential, that is, in the light of functional requirements, interchangability and
probable manufacturing circumstances.'' [17] Tolerances constrain a feature to lie within regions
known as tolerance zones and may relate to a single feature, for example a diameter, or to several
features, for example a distance between two holes. Features usually are referenced to datum
features which are used to establish the location and orientation of component datum's or
reference surfaces.

9
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

3.2.2. Geometric Dimension & Tolerance Data in CAD

The automation of many activities such as process planning and NC machining cannot be
successfully achieved based solely upon ideal geometry. Therefore a major requirement for
successful development and implementation of computer integrated manufacturing (CIM)
applications is a product model which includes allowed manufacturing inaccuracies, tolerances,
associated with the location and shape of component features.
Whilst 3D CAD modellers allow nominal geometry to be defined, they do not usually support
geometric dimensional & tolerance (GD & T) information. This is a known deficiency which is a
major problem when integrating systems which require tolerance data. [18] As GD&T
information must be associated with component surfaces in the geometric model it follows that
the user should operate through a common user interface to both the geometric model and the
GD&T model. [19, 20] This seemingly trivial requirement is not without its problems. [21]
3.2.3. Preferred Geometric Representations for Tolerancing of CAD Models.
Component geometry presented in a two dimensional drawing is open to interpretation and
cannot be relied upon to give an unambiguous representation of a product. It is therefore quite
possible that the original design intent can be lost. [22] Wireframe models lack mathematically
rigor therefore a surface or solid model based upon a boundary representation ( B-rep ) and/or
constructive solid geometry (CSG) are required as a basis of a product data modeller which
incorporates tolerances. [20,19]
If the underlying solid modeller uses boundary representation, linking tolerances to surfaces of
the model presents few problems. However, with CSG modellers specific surfaces are not
known to exist until the model is 'evaluated' in which case primitive surfaces may be split or even
totally disappear. [21]
3.2.4. Tolerance Representation in Research CAD Systems
Roy and Liu [12] developed a structured face adjacency graph (SFAG) over a hybrid CSG/B-rep
modeller. The face adjacency graph linked B-rep surfaces to CSG primitives in such a way that
all kinds of tolerance could be attached to the part model.
Gossard, Zuffante and Sakurai [23] used a combined CSG/B-rep modeller in a graph structure
known as an object graph. This stored dimensions as a relative position operator (RPO) which
described the location of a feature face relative to a reference face.
Requicha and Chan [24] presented a variational graph or VGraph tolerance representation based
upon CSG modeller PADL-2. The VGraph stores all tolerance information about an object
linked to a traditional CSG tree through a set of nominal faces (NFaces) of the object.
Wang and Ozsoy [25] proposed a hybrid CSG/B-Rep structure in which the CSG tree was
modified with the addition of constraining operator nodes (Cnodes) which constrains the
position or orientation between two primitives.
Kanai, Kawamura, Kishinami and Saito [26] used a hyper-patch model which was toleranced
through the application of Entity Attribute Relationships (EAR’s). These were also used to
attach information about machining processes used to generate the referenced surfaces.
EPS-1 [20, 27, 28] used two separate databases, one for dimensional and tolerance data and one
containing a B-rep model for nominal geometry. These two databases were accessed via a
common programming interface which allowed calls to be made to the underlying data structure
via an Application Interface Specification. The link between these two databases was a
dimension and tolerance (D&T) node consisting of a datum reference frame, which pointed to
the set of entities (features or surfaces) in the geometric model which were used as a reference, an

10
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

entity linking node, pointing to the specific entity on the model which was being toleranced and
an evaluated data node used to hold dimensions.
Walker and Wallis [29] used a CSG modeller which allowed attributes to be attached to surfaces.
Yau and Menq [18, 30, 31, 32] used CATIA's Graphic Interactive Interface (GII) and the
ATTRIBUTE function as a basis for associating tolerances with particular component surfaces.
Smolky and Vrana [22] described software which verifies that the model has been correctly
toleranced to ANSI Y14.5M and from this develop a worst case part or Softgauge® which is a
displayed along side the original geometry in the CAD system. This can be later used for
generation of CMM part programmes and for validation of the component. This is also described
by Foundyller [33] and Granquist. [34 ]
3.2.5. Tolerance Representation in Feature Based Systems
The use of form features or identification of features in an existing model (CSG or B-rep) offer a
way of linking manufacturing data to the CAD model [35] However, the use of design features
as a basis for subsequent or concurrent manufacturing process planning is not without its
difficulties. What constitutes a 'feature' will differ in design, manufacturing, assembly or
inspection contexts; the so called multiple-view problem. [13]. Also, a given component can be
generated, or decomposed, into an almost infinite number of ways. In practice only a
significantly small sub-set of these would be used but even this sub-set will not necessarily form
a good basis for automation of down stream activities, for example manufacturing process
planning or inspection process planning. [36] There is thus the need for 'feature' recognition,
feature refinement, feature conversion or feature mapping where feature is a context dependent
entity. All these are active areas of research interest. [13]
ElMaraghy and Gu [37] developed a feature based system offering a limited number of
toleranced primitives with known engineering and dimensional attributes. The selection of
features available representing a trade off between the flexibility and generality of the modelling
system developed and the complexity of the feature data structure.
Merat and Radack [38, 39] developed a feature base design system, the strategic design driven
inspector, based upon Concept ModellerTM [40] Associated with each feature type was a set of
alternative schema's for dimensioning and tolerancing the feature. When developing a
component, the user would create a starting feature, for example a block or cylinder, and add or
subtract additional features. When creating a specific feature, the actual dimensions and method
of tolerancing would be stated.

4. Component Inspection
4.1. Inspection Process Planning
4.1.1. Computer Aided Process Planning and CAD/CAM Systems.
Computer aided process planning (CAPP) systems are often regarded as being central to the
realisation of computer integrated manufacturing systems as they provide the link between
design activities and manufacturing activities. [19] Early CAPP systems were almost purely text
based and not typically associated with another planning activity, that of generating cutter paths
and the code necessary to drive computer numerically controlled (NC / CNC) machines. [41]
Due to the common link with computer aided design, these machining related systems are known
as CAD/CAM systems. In this article, the term Inspection Process Planning will be used to
encapsulate the high levels of automation typically associated with CAPP and the generation of
code for computer controlled machines or CAD/CAM.

11
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

4.1.2. Inspection Planning and the Programming of CMM’s

To inspect a component requires knowledgeable interpretation of design intent presented in
engineering drawings as well as a knowledge the processes used in its manufacture and of the
capabilities of the inspection available. From this information and knowledge a method to
inspect a component, which might not necessarily be on a CMM, can be developed. An
inspection process plan will include the sequence of inspection of datum surfaces and individual
features and, in the case of CMM’s, selection of suitable component and probe orientations and
the placing of probing points on the feature surfaces. [42] Inspection planning will also
determine how the raw inspection data is to be evaluated and how these results are to be
presented. An inspection plan may not necessarily be a formal document, but, in the case of
CMM’s would more commonly be embedded in the program used to drive a CMM. These
programmes are still often 'taught in' to the CMM; an initial manually driven inspection routine
being replayed for subsequent components. This programming overhead can be significant,
especially when relatively few parts are inspected.
4.1.3. CMM and CAD Interface Standards
The transfer of CAD geometry to off-line programming systems is usually made by either
bespoke routines or via standard CAD exchange standards such as IGES and VDA[43]. The
major problem with these interface standards is that they do not encompass tolerance data which
is an important data item for both the evaluation of inspection results, which is often done within
the CMM software, and also the automation of many inspection planning tasks.
CMM’s typically have their own programming language to which the off-line programming
system must adhere. The Dimensional Measuring Interface Specification (DMIS) is a standard
format used for bi-directional transfer of inspection data between CAD systems and CMM’s.
[44] DMIS is a high level programming language with a syntax similar to Automatically
Programmed Tool (APT). DMIS format part programmes and returned inspection results are
usually pre- and post-processed into a format usable by the CMM or CAD system. The
advantages of such standard interfaces include the demonstrated need and commitment of those
companies involved in its development. In the case of DMIS, numerous CMM suppliers and
users were involved. [44, 45] The emerging standard for the exchange of product data (STEP) is
likely to encompass geometric and process planning requirements in the future. However, the
development and acceptance of STEP. [46]
4.1.4. Off-Line Programming of CMM’s
Excluding metal removal, CMM programming is very similar to NC programming, both ideally
building upon a common base, a three dimensional model of the part. [20] However, whilst the
majority of expensive machining centres are today programmed either off-line or at the machine
whilst it is running another part program, CMM’s are just starting to follow this path.
Programming software for both off- and on-line programming of CMM’s is said to lie a decade
behind comparable software for NC machining centres [47] although this situation is changing
rapidly. A recent Delphi survey identified off-line programming tools as a major area of
development for the 1990's. [48]
4.1.5. Commercial Off-Line Programming Systems
Due to the similarity between programming of CNC machines and CMM’s, many CAD/CAM
vendors already offer off-line CMM programming capability, as do vendors of some robot
simulation systems. In addition, many CMM vendors have developed their own off-line
programming software. The functionality of these systems varies considerably, the more
advanced systems being totally integrated with a host CAD system and able to use the original
CAD geometry directly without the need for this to be transferred to the off-line programming
system via. a bespoke routine or a standard such as IGES or VDA.

12
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

Commercial offerings are typically a combination of an established CAD system coupled with a
third party inspection program. Available off-line programming systems for CMM’s include:
Prime/Calma, [49] EDS/Unigraphics, IBM's CATIA with Audi AG's Audimess or Valisys, [33,
34] Matra Datavision's Euclid with GEMINI and CAMPS 3D, Silma with CimStation Inspection,
[47] Tecnomatixs with Robcad/CMMxWorks, [50] Brown & Sharp with Microquindos, [51]
Qualstar/First [49] and Mitutoyo with CMMCAD. Bespoke developments include Lehr
Precision integrating with McDonnell Douglas [52] and Mitutoyo/Geopak with a number of CAD
systems. [1]
4.2. Research Inspection Process Planning Systems
Few fully automated inspection planning systems have been realised, although all research into
inspection process planning contain elements of automation. Those research groups that have
reported complete systems, as opposed to concentrating upon elements of the process planning
task, break the problem of process planning down into different stages.
4.2.1. Expert Planning System One - EPS-1
Expert Planning System One (EPS-1) is a research system developed by CAM-I. This testbed
system aimed to develop an intelligent inspection process planning environment leading to highly
automated dimensional inspection planning through the integration of CAD, CAM and CAPP.
EPS-1 included a geometric modeller, a dimensioning and tolerancing modeller and an inspection
process plan generator. Considerable use of built in logic aimed to automate the process as much
as possible. Automated facilities included selection of appropriate CMM, fixturing, global
inspection plans and probing sequences for individual features. [19, 20, 27]
EPS-1 breaks the problem of inspection process planning down into nine parts:
a) obtain/define inspection plan.
Definition of part id's and inspection strategy constraints.
b) task decomposition/definition.
Extraction of toleranced features and datum surfaces from the GD&T modeller. For each
feature a 'work element' is added to the overall inspection plan.
c) determine methods.
Selection of a suitable CMM based upon the types of features to be measured.
d) determine set-up
Optimisation orientation of the part with respect to the CMM using orientation vectors for
each feature. Work element accessibility is based upon machine axis, feature surface
normal's or cylindrical axis and reference / datum feature orientation. The rules used
minimised the total number of component orientations required.
e) determine tool/holder
Probe orientations required are selected for each feature / work element heuristically.
Strategy may be to minimise the number of probes used or to maximise the number of
probing's from a given probe.
f) detail/optimise plan
Sequencing of inspection activities / work elements is determined via heuristics and
decision table logic. Datum and feature precedence is accounted for.
g) generate/simulate
Individual feature probing sequences are generated using templates, and the path of viewed
as a 3D line in the geometric modeller. Clash detection and avoidance must be carried out
manually as no clash detection was available.
h) produce control information
Produce DMIS output for the measurement task.
i) produce support information
Add further information required by the operator.

13
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

4.2.2. Inspection Process Planning Expert (IPPEX)

Brown and Gyrog [19] developed Inspection Process Planning Expert (IPPEX) as a further
refinement of the concepts demonstrated by EPS-1.
4.2.3. Intelligent Planning Environment
Yau and Menq [18, 30, 31, 32] implemented an automated inspection planning system based
upon the commercial CAD system CATIA and a knowledge based expert system shell. This
system suggested five steps in inspection planning;
1. Specification
Translate product data, geometry, tolerances, manufacturing methods, inspection
constraints etc. into a specification of what is to be measured and how.
2. Inspection Planning
Locate probing points, defining approach vectors, plan moves between individual probing
points and between features.
3. Program Verification
Check this initial path for clash situations and if necessary corrected.
4. Execution
Execute this verified program.
5. Analysis
Analyse the resulting data points against the nominal geometry and generate an inspection
report.
4.2.4. McMaster Inspection Planner
ElMaraghy and Gu [37] used an expert system written in Prolog coupled with a research
geometric / GD&T modeller.
1. Feature identification
Selects those features which are appropriate for inspection on a CMM
2. Accessibility Planning
Identifies those features / datum's which can be inspected with the current
probe/component orientation,
3. Sequence Planning
Sequences datum inspection followed by inspection of features related to these datum's.
4. Tolerance Checking
Provides feedback of inspection results.
4.2.5. Strategic Design Driven Inspector (SDDI)
Merat, with Radack and Galm [38, 39, 53] described a Strategic Design Driven Inspection system
which uses 'fast' but limited accuracy techniques such as vision based systems as a method of
eliminating features with relatively course tolerances or which are obviously outside tolerance,
before using slower instruments such as a CMM and contact probe.
1. Generate Inspection Plan Fragments
Each feature or datum surface is linked to an Inspection Plan Fragment Generator for that
particular class of tolerance. The IPFG produces Inspection Plan Fragments which contain
valid probe / component orientations for the particular feature.
2. Group measurement requests by part orientation
All IPF's which can be measured with the same part orientation are grouped together.
3. Select an IPF to inspect each feature
One IPF from those generated by the IPFG is selected based upon optimisation rules.
4. Eliminate redundant measurement requests
IPF's for different tolerances can generate measurement requests for the same surface.
Duplicate requests are removed.

14
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

5. Create measurement sequences from measurement requests

Each IPF is used to generate a sequence of probe movements.
6. Sequence the operations
A valid sequence, taking account of precedence etc. is generated using a topological sort
algorithm.
7. Generate collision free path
Configuration spaces are used to generate movements between individual features.
8. Generate executable code
Executable code in DMIS format is generated.
4.2.6. Stuttgart IPA
Roth-Koch [54] described a system under development at Stuttgart IPA for the inspection of free
form surfaces and evaluation of inspected data based upon the CATIA CAD system.
4.2.7. Brunel CMM-CAD
Mullineux with Medland, Singh, Sittas and also with Cowling [55, 56, 57] described a constraint
based system under development at Brunel University.
4.2.8. New South Wales PC Based Inspection
Farmer and Smith [58] described a PC based inspection planning system where inspection data
coded into feature attributes in a 2D AutoCAD drawing were used to drive a CMM.
4.3. Elements of Inspection Process Planning
4.3.1. Component / Probe Orientation Strategy.
Numerous valid orientations of a component / probe are possible. An important part of
inspection planning is to establish which component orientations are required to allow inspection
of all features. These orientations must take account of accessibility of features and datum
surfaces, and also machine axis and possible probe orientations.
Spyridi and Requicha [59] developed the idea of feature accessibility cones. This abstracted the
actual probe as a straight line and for each feature determined those probe orientations which
allowed contact with the entire feature surface. The set of possible orientations was termed the
Local Accessibility Cone (LAC). In order to account for situations where valid probe
orientations intersected the workpiece, i.e. a clash situation, a Global Accessibility Cone (GAC)
was also determined which was the set of probe orientations which did not pass through the
workpiece. Given that a set of features can be inspected by the same probe if there is a none
empty intersection of their GAC's, determining the minimum number of probe orientations
required to inspect all features is a matter of finding the minimum number of sets of features with
none empty GAC intersections. Finding the optimal minimum number of sets is an insoluble
problem, but algorithms for determining sub-optimal sets are possible. Other techniques must be
used to select appropriate orientations from the candidate orientations suggested.
Khoshnevis and Yeh [60] developed the similar concept of accessibility zones for given feature
surfaces. An accessibility zone was the area of the feature accessible given the available probe
orientations.
4.3.2. Probe Point Placement Algorithms.
An important aspect of inspection process planning is placement of probing points on a
candidate feature. The number and location of inspection points should reflect both the
geometry of a feature and the tolerance applied to it. When a 'hard' gauge is replaced by a CMM,
the nature of the inspection changes. Inspection using a CMM only samples the surface at a
finite number of points. This leads to two related problems; where to sample the surface and
how to interpret the resultant data. [38]

15
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

Mullineux with Medland, Singh, Sittas and also with Cowling [55, 56, 57] used standard probing
algorithms for regular features and point to point inspection for other, none regular surfaces.
Standard algorithms included equi-distant points on the circumference of a circle, the number of
points to be inspected could be defined interactively.
EPS-1 [19, 20, 27] used a template for each feature assuming an ideal relative orientation of probe
and feature. This idea was extended by Legge and Åhman [61] to compensate for relative probe /
feature orientation. Merat and Radack [38, 39] used a template known as an Inspection Plan
Fragments Generator (IPFG) which produced inspection code fragments, i.e. process plans for
individual feature, depending upon tolerance feature type.
Walker and Wallis [29] used a regular rectangular grid projected upon the surface of the feature
to be measured, placing probing points at a random positions inside each grid cell. These points
were then grouped into 'clusters', the cluster size (and grid mesh) depending upon feature type
and tolerance. Each cluster was subsequently replaced by a representative probing point.
Vardar and Ozsoy [62] used a method for the placement of probing points on a planar surface
which was derived from the technique of mesh generation for Finite Element Analysis (FEA). A
candidate set of inspection points was given by the centroid of each of the triangular mesh
patches from the mesh generation algorithms. Three methods of selecting the required number
of points from this candidate set were presented; random selection, maximising the distance
between selected points and selection of the outermost points within the meshed surface.
Kanai, Kawamura, Kishinami and Saito [26] used an integrated CMM / NC programming system
to locate probing points in such a way as to detect expected inaccuracies from the machining
process, for example surface waviness when face milling.
Yau and Menq [18, 30, 31, 32] used a statistical sampling plan to determine the number of
probing points required based upon the design tolerance and the accuracy of the manufacturing
processed used to generate the feature. This information was held as Inspection Attributes
associated with each feature in the CAD system used. The sampling plan gave the number of
inspection points but not their distribution over the feature and, as would be expected, increased
the number of sample points in response to tighter tolerances, but reduces them as process
capability improves.
Duffie, Feng and Kann [11] generated paths for probing complex sculpted surfaces which drove
the probe normal to the surface, with a given scanning density. This technique was also
described by Roth-Koch [54] and Reynolds [52] Kawabe, Kimura and Sata [63] increased the
probing density at the edge free form surface to reduce the influence of a zero curvature
assumption at the boundary. Mullineux with Medland, Singh, Sittas and also with Cowling [55,
56, 57] suggest a similar procedure in which probe points along two adjacent features are made
closer together towards the expected/predicted feature transition.
Chen, Tang, Ni and Wu [64] described predictive algorithms for placement of probing points
when scanning free form surfaces. These algorithms increased the scanning density where the
surfaces experienced large rates of change. The same technique was used by Jennings [49]
4.3.3. Sequence of Probing
The probing points on a given feature can usually be probed in any sequence. Likewise, the
sequence of inspection of features is only constrained by the requirement to inspect datum
features first. It is therefore possible to minimise the overall inspection time by selecting an
optimum sequence of execution of features and feature probing sequences.
Merat and Radack [38] determine the sequence of probing for the set of points related to a feature
by starting from an arbitrary point and moving to the nearest unprobed point on the current

16
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

feature. Walker and Wallis [29] introduce the idea of a cost function proportional to the weighted
distance between the points being considered; the weighting accounts for different axis of the
CMM taking different times to move the same distance. The same technique is used to select the
next feature to inspect.
4.3.4. Clash Avoidance
A CMM can be regarded as a Cartesian robot whose end effector is the probe tip. The aim of
inspection process planning is to generate a collision free path through all inspection points. Two
possible methodologies are possible; clash avoidance and clash detection / evasion. In clash
avoidance schemes, clash situations are avoided when defining a probe path. In clash evasion
schemes candidate probe paths are evaluated for clash situations which, if found, are corrected.
[65] In both situations it is important to know the distance between objects during relative
movement as well as the extent of a clash if it should occur. Dong and Yuan [66] identify a
variety of techniques including vertex based, surface based and convex object based methods for
calculating the distance between object.
Merat and Radack [38] used pre-calculated configuration spaces for given feature orientations
and CMM / probe combinations. Using the fact that the configuration space for a component is
the union of the configuration spaces of its individual features, were able to generate a
component configuration space. This was then used with a minimum path algorithm to derive
allowable paths within the configuration space.
Yau and Menq [18, 30, 31, 32] consider the avoidance of clash by selection of alternative probe
orientations. The idea of a local accessibility cone developed by Spyridi and Requicha [59] were
modified to take account of only those points being inspected rather than the entire surface of the
feature. From the candidate orientations found, the probe orientations which allow access to the
inspection points can be determined. A heuristic method was used to select a probe orientation
that does not clash with the component during approach to these points.
4.3.5. Clash Detection and Evasion
Walker and Wallis [29] approached the problem of clash detection by using the technique of
shrinking the model of the probe tip (or any other part of the probe model ) to a point and
growing the model of the component being inspected by a corresponding amount. If the probe
tip centre lies inside the grown component at any time a clash situation exists. This is a technique
well known in the robot path planning community. [66, 67] This clash was subsequently avoided
by using a heuristic search technique to locate a point on the edge of the surface to be avoided
which was closest to the destination point. A transit point is created which is offset from this
edge point.
Yau and Menq [18, 30, 31, 32] used a heuristic method of clash detection and correction. In this
work, the probe tip was taken to be a point and the CMM was assumed to move with straight line
path segments. A clash situation occurred if the line representing the movement of the tip
intersected the component surface. This idea was extended to include modelling of the stem of
the CMM probe as a line, (which becomes a face when the stylus moves along a straight line
path) and the column as a cylinder, (with a swept volume approximated to a box). A clash
situation was known to have occurred when the stem 'face' or any face of the column 'box'
intersects the component. A heuristic method was then used to modify the probe path by the
placement of additional transit points if a clash situation was discovered however no details were
given.

17
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

4.3.6. Generation of DMIS Programmes

A final stage is for the validated inspection paths to be output in a format acceptable for the
CMM to be used. This format could be the CMM’s native programming language or the
dimensional measuring interface standard (DMIS).
Pham, Martin and Khoo [68] described a knowledge based expert system to provide a generic
link from inspection planning systems to a CMM. Through a dialogue with the user, the expert
system was given information relating to the format of the input data file and subsequently
generated a pre-processor to convert an inspection plan into Neutral Data Format (NDF - an
equivalent of DMIS.)
4.3.7. Scanning and Digitisation of Free Form Surfaces.
Scanning of a surface where the actual geometry is not known is a special case where the
inspection process plan must be created dynamically.
Chen, Tang, Ni and Wu. [64] noted two conditions for accurate digitisation: a.) that the
measuring probe contacts normal to the component surface and b.) that a sufficient number of
measurement points are taken. Two ways to achieve the first were reported, the use of probes
that detect the contact point and the use of the original CAD model to generate the probe path.
Three methods for scanning a free form curve were presented; Linear Prediction, Polynomial
Prediction and Spline interpolation. These methods used on-line measured data to predict the
form, and hence contact points and normal, of the surface being scanned. The sampling rate
being adjusted dynamically based upon actual changes in the curvature of the curve.
4.4. Evaluation
4.4.1. Deterministic and Random Measurement Errors
In order to evaluate the raw data from the CMM, it has to be known that this raw data is in its self
accurate. Yau and Menq [18, 30, 31, 32] proposed a statistical method of separating random
errors from deterministic errors when measuring surfaces of a known form.
Deterministic errors can be compensated for by correcting the measured probe points with a
mathematical model of the CMM determined by either measuring a testpiece with known
geometry, for example a three dimensional ball plate, or by measuring individual geometric
parameters using an independent measurement technique such as laser interferometry. These are
known as artefact testing and parametric calibration respectively. A total of 21 parameters can be
identified: three scale errors, vertical and horizontal straightness and three angular errors per axis,
and three squareness errors. These techniques can improve accuracy by a factor of ten. [6, 69]
4.4.2. Conversion of Raw Data Points to Actual Geometry
A CMM, like many other methods of inspection, only generates samples of data points on an
individual component. There are currently no standards or obvious interpretations of this
measurement data, which is regarded by many as a significant problem. However, formal
algorithms for the evaluation of point data, based upon functional tolerancing rather than ease of
computation, are being developed. [70, 71, 72] It must also be noted that analysis software itself
is a potential source of error. [6]
4.4.3. Location of the Component on the CMM.
Yau and Menq [18, 30, 31, 32] reiterate the commonly used 3-2-1 method; three points on a
primary plane, two on a secondary plane and one on a tertiary plane. A 'localisation' algorithm
was then used which minimised the sum of the squared distances between a probed point at a
known location and the corresponding points on the CAD model and from this data derives a
transformation matrix. Further points were fitted to the model using this matrix. However,
although the component has been mathematically oriented to the original CAD geometry, this
localisation is itself influenced by manufacturing inaccuracies.

18
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

4.4.4. Fitting of Inspected Points to Regular Features

The use of least squares fitting of inspection data to nominal form features, lines, circles etc. is
extensively used. Yau and Menq [18, 30, 31, 32] propose an Optimal Match algorithm to
eliminate the offset error between measured and design data by minimising the squared shortest
distances from each point to the nominal feature geometry. The same technique was used by
Mullineux with Medland, Singh, Sittas and also with Cowling [55, 56, 57]
Jennings [49] also presented a least squares solution for mapping inspected data points onto
nominal geometry. His technique used the measured point and a point projected onto the
nominal feature surface; the measured point assuming to lie on a vector normal to this nominal
feature point. It was pointed out that by skewing the density of the measured data points in a
certain area, the fit would be skewed towards the more densely probed area. This problem was
eliminated by the development of a geometric tolerancing best fit whereby each feature surface
was assigned a tolerance which was converted into a tolerance band. The inspected data was
then allowed to 'float' until it sat within the tolerance band. A correction matrix for position and
alignment could then be produced. Duffie, Feng and Kann [11] reported 'best fitting' of sets of
data points and error compensation in applications such as mould design.
4.4.5. Curve Fitting
Kawabe, Kimura and Sata [63] used curve fitting, increasing the probing density at the ends of
the curve in order to reduce the influence of a zero curvature assumption at the boundary.
Reynolds [52] reports a system for feeding back inspection data from aerofoil sections to a CAD
system. Granquist [34] describes an environment where die casting moulds are inspected and
error reports fed back graphically to the CMM inspector. Smith [9] reports similar work with
injection moulding dies as does Duffie. [73] Gupta and Sagar [74] reported a system for the
capture and interpretation of 3D point data.
The geometry of a component feature generated by curve fitting may be significantly different to
that expected from the original CAD model. If this error is unacceptable Mullineux with
Medland, Singh, Sittas and also with Cowling [55, 56, 57] suggest a method of sub-division
where the inspected feature is divided into two and re-inspected. This procedure isolates the
geometric inaccuracies in the actual component into a specific part of the curve.
4.4.6. Graphical Feedback
Mullineux with Medland, Singh, Sittas and also with Cowling [55, 56, 57] assessed circular
features by comparison of nominal features with actual features; the error in the position and size
of the actual feature being scaled by a factor of 100. Jennings [49] used a similar technique,
providing graphical feedback of errors was made by scaled error vectors showing the deviation
from nominal position. This was also reported by Duffie, Feng and Kann. [11]
4.4.7. Evaluation of Source of Errors
Kanai, Kawamura, Kishinami and Saito [26] used Entity Attribute Relationships (EAR’s) to not
only attach tolerance information but also the machining processes used to generate the
referenced surfaces. If a feature was out of tolerance the EAR was used to identify the
manufacturing process which caused the error. A similar technique was presented by Danzer.
[75]

19
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

5. Conclusions and Future Developments

The techniques for automatic and semi-automatic off-line programming of co-ordinate measuring
machines have been well researched and continue to be incorporated into commercial systems.
This development will continue, probably at a pace driven by market demand rather than
availability of techniques.
A significant bottleneck to system development today is the availability of CAD systems that
allow tolerancing of 3D models and secondly the available of neutral interface formats for
transfer of this information; many of the more competent off-line programming systems have
bespoke links to market leading CAD systems.

6. Acknowledgements
As this article represents a review of published work it is impossible to include the most recent
developments. Limitations of space also preclude a more detailed analysis of individual aspects
of inspection process planning, which, after all, are dealt with most thoroughly in the original
works. I hope that the researchers and research groups whose work is mentioned here will bear
this in mind when criticising me for the poor coverage I have given their work.
Acknowledgement is duly given to my co-workers in the Department of Manufacturing
Engineering at Luleå University of Technology and especially to NUTEK for funding.

7. References
1 T. Inglesby, "CMM in CIM." Manufacturing Systems October 1989 pp18-25
2 J.A. Bosch, ''Planning Overview: Systems Aspects in Flexible Inspection for Automated
Manufacturing.'' ASME Manufacturing Review Vol.2 No.1 March 1989 pp26-31
3 J. Raja and U.P. Sheth, "Integration of Inspection into Automated Manufacturing
System." Recent Developments in Production Research 1988.
4 W. Tandler, "Critical Concepts of High Performance Co-ordinate Measuring Systems."
Multi-Metrics Incorporated, Redwood City, California, USA. 1984.
5 A. Sostar, "Co-ordinate Measuring Technique in Quality Assurance." Robotics and
Computer Integrated Manufacturing Vol.4 No.1/2 pp259-265
6 H. Kunzmann and F. Wäldele, "Performance of CMM’s." Annals of the CIRP Vol.37,
2/1988 pp633-640
7 G.L. Bowen and L.S. Duncan, "Integrated Metrology System." Booz, Allen & Hamilton
Incorporated, Cleveland Ohio, USA.
8 J. Bosch, "The Case for CMM’s." Tooling and Production October 1988.
9 L. Smith, "CMM Smoothes Mold Design at Ford." American Machinist, April 1993
10 D.L. Taylor "Computer Aided Design." ISBN 0 201 16891 X
11 N. Duffie, S. Feng & J. Kann, "CAD-Directed Inspection, Error Analysis and
Manufacturing Process Compensation Using Tricubic Solid Databases." Annals of the
CIRP Vol.37, 1/1988
12 U. Roy and C.R. Liu, "Feature Based Representational Scheme of a Solid Modeller for
Providing Dimensioning and Tolerancing Information." Robotics and Computer
Integrated Manufacturing 1988 Vol.4 No.3/4

20
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

13 O.W. Salomons, F.J.A.M. van Houten & H.J.J. Kals, "Review of Research in Feature-
Based Design." Journal of Manufacturing Systems Vol.12 No.2
14 R-K. Li and S. Adiga, "Part Feature Recognition - A vital Link in the Integration of CAD
and CAM." Recent Developments in Production Research 1988.
15 N.K. Shaw, M.S. Bloor & A. de Pennington, "Product Data Models." Research in
Engineering Design 1989/1 pp43-50
16 T. Tomiyama, T. Kiriyama, H. Takeda, D. Xue & H. Yoshikawa, "Mete-Model: A Key to
Intelligent CAD Systems." Research in Engineering Design 1989/1 pp19-34
17 "Dimensioning and Tolerancing." Swedish Standard SS ISO 1101
18 H-T. Yau and C-H. Menq, "An Automated Dimensional Inspection Environment For
Manufactured Parts Using Co-ordinate Measuring Machines." International Journal of
Production Engineering Research. 1992 Vol.30 No7 pp1517-1536 ISSN 0020 7543
19 C.W. Brown and D.A. Gyorog, "Generative Inspection Process Planner for Integrated
Production." 1990 Winter Annual Meeting of the American Society of Mechanical
Engineers. ASME PED Vol.47 pp151-162
20 L. Patrick, "EPS-1 An Expert Programming System For Dimensional Measuring
Equipment Programming." 24th Annual Meeting and Technical Conference Proceedings -
AIM Tech '87: Back To Basics. ISSN 0882 5548
21 N.P. Juster "Modelling and Representation of Dimensions and Tolerances." Computer
Aided Design, Vol.24, No.1 January 1992.
22 R. Smolky and J.J. Vrana, "CMM: In-Process Inspection Analysis on a CAD System."
Worldwide Passenger Car Conference. ISBN 1 56091 296 0
23 D.C. Gossard, R.P. Zuffante & H. Sakurai, ''Representing Dimensions, Tolerances and
Features in MCAE Systems.'' IEEE Computer Graphics and Applications. March 1988
24 A.A.G. Requicha and S.C. Chan, "Representation of Geometric Features, Tolerances and
Attributes in Solid Models Based On Constructive Geometry." IEEE Journal of
Robotics and Automation Vol.2 No.3 September 1986
25 N. Wang and T.M. Ozsoy, "A Scheme to Represent Features, Dimensions and
Tolerances in Geometric Modelling." Journal of Manufacturing Systems, Vol.10, No.3
26 S. Kanai, T. Kishinami, R. Kawamura & K. Saito, "The Computer Aided Testing and
Diagnostic Systems of the Manufacturing Process using the Co-ordinate Measuring
Machine." Proceedings of NAMRC XVII pp.311-318 ISBN 0-872-63-356-X
27 M.D. Reimann and J. Sarkis, "An Architecture for Integrated Automated Quality
Control." Journal of Manufacturing Systems. Vol.12 No.4.
28 P.S. Ranyak and R. Fridshal, ''Features For Tolerancing a Solid Model.'' Proceedings of
the 1988 ASME International Computers in Engineering Conference and Exposition.
29 I. Walker and A.F. Wallis, "Application of 3D Solid Modelling to Co-ordinate Measuring
Inspection." Proceedings of the 5th. International Conference on Metrology and Properties
of Engineering Surfaces. ISSN 0890 6955
30 C-H. Menq, H-T. Yau & C-L. Wong, "An Intelligent Planning Environment for
Automated Dimensional Inspection using Co-ordinate Measuring Machines." Journal
of Engineering for Industry. ISSN 0022 0817.

21
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

31 H-T. Yau and C-H. Menq, "Path Planning for Automated Dimensional Inspection Using
Co-ordinate Measuring Machines." Proceedings of the 1991 International Conference on
Robotics and Automation. Vol.3 pp1934-1939
32 C.H. Menq, C.L. Wong & H.T. Yau, ''An Intelligent Planning Environment for
Automated Dimensional Inspection of Manufactured Objects.'' Ohio State University.
33 C. Foundyller, ''Tolerancing: Common Thread of MCAE.'' CAE August 1988.
34 J. Granquist, "Closing the Loop of Mold Design, Manufacturing and Inspection."
American Machinist April 1993.
35 H.D. Park and O.R. Mitchell, "CAD Based Planning and Execution of Inspection."
Proceedings of the 1988 Computer Society Conference of Computer Vision and Pattern
Recognition. pp858-863
36 K.E. Hummel and C.W. Brown, ''The Role of Features in the Implementation of
Concurrent Product and Process Design.'' Proceedings from Symposium on Concurrent
Product and Process Design, AMSE Winter Meeting 1989. pp1-8
37 H.A. ElMaraghy and P.H. Gu, "Expert System for Inspection Planning." Annals of the
CIRP Vol.36, 1/1987
38 F.L. Merat, G.M. Radack, K. Roumina & S. Ruegsegger, "Automated Inspection Planning
within the Rapid Design System." IEEE CH 3051-0/91/0000-0042
39 F.L. Merat and G.M. Radack, "Automatic Inspection Planning Within a Feature Based
CAD System." Robotics and Computer Integrated Manufacturing. Vol.9 No.1 pp61-69
ISBN 0 7803 0173 0
40 Wisdom Systems, Pepper Pike, Ohio, USA
41 W. Eversheim and J. Scheenwind, ''Computer Aided Process Planning -- State of the Art
and Future Development.'' Robotics and Computer Integrated Manufacturing 1993 Vol.10
No.1/2 pp65-70
42 T.H. Hopp, "CAD-Directed Inspection." Annals of the CIRP Vol.33 1/1984
43 ''IGES and Beyond.'' CIM Technology, Summer 1985 pp20-21
44 ''The Dimensional Measuring Interface Standard.'' ANSI CAM-I 101, 1990
45 J. Zink, ''Linking CAD/CAM Systems to CMM’s.'' Proceedings of Test, Measurement and
Inspection for Quality Control. Detroit 1987
46 B.J. King and P.W. Norman, "A STEP in the Right Direction." Professional Engineering,
November 1992
47 F. Mason, "Program Your CMM Off-Line." American Machinist October 1992.
48 1991 Report by Future Technology Surveys Incorporated, Lilburn, GA, USA.
49 R.M. Jennings, ''Feature Recognition Reduces CMM Programming Time Whilst
Speeding Analysis.'' SME Technical Paper MS91-293 119
50 "CMM Support Systems Speed Throughput." Manufacturing Engineering March 1992.
51 "Fine-Tuning Quality." Manufacturing Engineering November 1992.
52 "CAD to CMM Link Produces Precise Airfoils." CAE June 1988

22
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

53 J.G. Galm and F.L. Merat, ''The Strategic Design Driven Inspection of Machined Parts.''
1988 IEEE Conference on Computer Integrated Manufacturing
ISBN 0-816-0888-9
54 S. Roth-Koch, ''Measuring as a CAD Application.'' Interfaces in Industrial Systems for
Production and Engineering 1993.
55 A.J. Medland, G. Mullineux, R. Singh & E. Sittas, "A Modular Approach to Linking
Computer Aided Design and Automatic Inspection Systems." Department of
Manufacturing Systems, Brunel University, Uxbridge, UK.
56 G.J. Cowling and G. Mullineux, "Toward an Intelligent CAD-CMM Interface."
Engineering with Computers 5. pp133 - 141
57 A.J. Medland, G. Mullineux, ''A Constraint Approach to Feature-Based Design.''
International Journal of Computer Integrated Manufacturing, Vol.6 No's.1&2, 1993, pp34-38
58 L.E. Farmer and G. Smith, "Integrating CAD and CMM Inspection." Proceedings of the
Australian Institute of Engineers International Mechanical Engineering Congress and
Exhibition - MECH 91. ISSN 0313 6922
59 A.J. Spyridi and A.A.G. Requicha, "Accessibility Analysis for the Automated Inspection
of Mechanical Parts by Co-ordinate Measuring Machines." Proceedings of the 1990 IEE
Conference on Robotics and Automation. ISBN 0 816 2061 7
60 B. Khoshnevis, Z. Yeh, ''Automatic Measurement Planning For Co-ordinate Measuring
Machines.'' Proceedings of the 8th International Conference on CAD/CAM, Robotics and
Factories of the Future. pp581-590
61 D.I. Legge, P. Åhman, ''Feature Probing Algorithms Used in a Semi-Automatic CMM
Programming System.''
62 H.O. Vardar and T. M. Ozsoy, "A Method for Selecting Measurement Points On Planar
Faces of Objects with Holes." Computers in Engineering. Vol.1 pp429-436 ISBN 0 7918
0935 8
63 S. Kawabe, F. Kimura & T. Sata, "Generation of NC Commands for Sculptured Surface
Machining from 3-Coordinate Measuring Data." Annals of the CIRP Vol.29, 1/1980
64 Y.D. Chen, X.J. Tang, J. Ni & S.M. Wu, "Automatic Digitisation of Free Form Curves By
Co-ordinate Measuring Machines." Proceedings of the Winter Annual Meeting of the
American Society of Mechanical Engineers. ISBN 0 7918 1124 7
65 S.M. Udupa, ''Collision Detection and Avoidance in Computer Controlled
Manipulators.'' Proceedings of the 5th. MIT International Conference on Artificial
Intelligence.
66 Z. Dong and J. Yuan, "A Formulation for Collision Identification and Distance
Calculation in Motion Planning Using Neural Networks." International Journal of
Advanced Manufacturing Technology, 1993-8 pp227 - 234
67 T. Lozano-Perez and M.A. Wesley, "An Algorithm for Planning Collision Free Paths
Among Polyhedral Obstacles." Communications of the ACM October 1979 Vol.22 No.10
68 D.T. Pham, K.F. Martin & L.P. Khoo, "A Knowledge Based Pre-Processor Generator for
Co-ordinate Measuring Machines." International Journal of Production Research 1991
Vol.29 No.4 pp677-694 ISSN 0020 7543

23
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

69 A.K. Elshennawy and I. Ham, "Performance Improvement in Co-ordinate Measuring

Machines by Error Compensation." Journal of Manufacturing Systems Vol.9 No.2
70 F. Etesami and H. Qiao, "Analysis of Two-Dimensional Measurement Data for
Automated Inspection." Journal of Manufacturing Systems Vol.9 No.1
71 G. Caskey, Y. Hari, R. Hocken, D. Palanvelu, J. Raja, R. Wilson, K. Chen & J. Yang,
''Sampling Techniques for Co-ordinate Measuring Machines.'' Proceedings of the 1991
NSF Design and Manufacturing Systems Conference, Austin, Texas
72 X. Zhang and U. Roy, "Criteria for Establishing Datum's in Manufactured Parts."
Journal of Manufacturing Systems Vol.12 No.1
73 N.A. Duffie, "Integration of Automated Inspection and Database Modification
Capabilities into Manufacturing Systems." University of Wisconsin-Madison, USA
74 V.K. Gupta and R. Sagar, "A PC Based System Integrating CMM and CAD for
Automated Inspection and Reverse Engineering." International Journal of Advanced
Manufacturing Technology 1993-8 pp305 - 310
75 H.H. Danzer, "Application of 3-Dimensional Co-ordinate Measuring Machines for
Problem Investigation and Upstream Quality Assurance." Annals of the CIR Vol.36,
1/1987

8. Alphabetical List of Authors

S. Adiga 14 D.A. Gyorog 19
M.S. Bloor 15 I. Ham 69
J.A Bosch 2, 8 Y. Hari 71
G.L. Bowen 7 R. Hocken 71
C.W. Brown 19, 36 T.H. Hopp 42
G. Casky 71 F.J.A.M. van Houten 13
S.C. Chan 24 K.E. Hummel 36
K. Chen 71 T. Inglesby 1
Y.D. Chen 64 R.M. Jennings 49
G.J. Cowling 56 N.P. Juster 21
H.H. Danzer 75 J. Kann 11
Z. Dong 66 S. Kawabe 63
N. Duffie 11, 73 L.P. Khoo 68
L.S. Duncan 7 B. Khoshnevis 60
H.A. ElMaraghy 37 F.Kimura 63
A.K. Elshennawy 69 B.J. King 46
F. Etesami 70 T. Kiriyama 16
W. Eversheim 41 H.J.J. Kals 13
L.E. Farmer 58 S. Kanai 26
S. Feng 11 R. Kawamura 26
C. Foundyller 33 T. Kishinami 26
R. Fridshal 28 H. Kunzmann 6
P.H. Gu 37 D.I. Legge 61
J.G. Galm 53 R-K. Li 14
D.C. Gossard 23 C.R. Liu 12
J. Granquist 34 T. Lozano-Perez 67
V.K. Gupta 74 K.F. Martin 68

24
 Paper 1: Integration of Design and Inspection Systems - A Research Review.

F. Mason 47 I. Walker 29
A.J. Medland 55, 57 A.F. Wallis 29
O.R. Mitchell 35 N. Wang 25
C-H. Menq 18, 30, 31, 32 M.A. Wesley 67
F.L. Merat 38, 39, 53 R. Wilson 71
G. Mullineux 55, 56, 57 C-L. Wong 30, 32
J. Ni 64 S.M. Wu 64
P.W. Norman 46 F. Wäldele 6
T.M. Ozsoy 25, 62 D. Xue 16
D. Palanvelu 71 J. Yang 71
H.D. Park 35 H-T. Yau 18, 30, 31, 32
L. Patrick 20 Z. Yeh 60
A. de Pennington 15 H. Yoshikawa 16
D.T. Pham 68 J. Yuan 66
H. Qiao 70 X. Zhang 72
G.M. Radack 38, 39 J. Zink 45
J. Raja 3, 71 R.P. Zuffante 23
P.S. Ranyak 28 P.Åhman 61
M.D. Reimann 27
A.A.G. Requicha 24, 59
S. Roth-Koch 54
K. Roumina 38
U. Roy 12, 72
S. Ruegsegger 38
R. Sagar 74
K. Saito 26
H. Sakurai 23
O.W. Salomons 13
J. Sarkis 27
T. Sata 63
J. Scheenwind 41
N.K. Shaw 15
U.P. Sheth 3
R. Singh 55
E. Sittas 55
L. Smith 9
G. Smith 58
R. Smolky 22
A. Sostar 5
A.J. Spyridi 59
H. Takeda 16
W. Tandler 4
X.J. Tang 64
D.L. Taylor 10
T. Tomiyama 16
S.M. Udupa 65
H.O. Vardar 62
J.J. Vrana 22

25
 Paper 2

Semi-Automatic Probe Path Planning for

Feature Based CMM Programming.

26
 Paper 2: Semi-Automatic Probe Path Planning for Feature Based CMM Programming.

Semi-Automatic Probe Path Planning for

Feature Based CMM Programming.
Eur. Ing. David Ian Legge
Department of Manufacturing Engineering,
Luleå University of Technology, Sweden
Abstract
This article describes algorithms used for the automatic generation of feature probing sequences
in an off-line programming system for co-ordinate measuring machines (CMM’s). These
algorithms were formulated during development of a prototype inspection planning module for a
commercial off-line programming system originally developed for the programming of robots.
The algorithms form a basis for the automation of inspection planning tasks which are tedious to
perform, whilst leaving the operator free to apply inspection methodology to the overall
inspection task. Inspection planning in the system developed is therefore semi-automatic.
The algorithms presented address the placement of probing points on a feature surface, predictive
modification of the probing pattern to prevent clash due to the relative orientation of the
probe/feature and clash avoiding probe path planning. Clash evasion when traversing between
features is also considered.
Keywords
Co-ordinate Measuring Machines, Inspection, Off-Line Programming, Features.

Background
The Need For Off-Line Programming Tools.
The benefits of Co-ordinate Measuring Machines (CMM’s) are well recognised [1]. The high
productivity of Numerically Controlled (CNC) machines requires near instant component
verification; indeed the purpose of a co-ordinate measuring machine is often to evaluate and
monitor the performance of a machine tool [2]. For this verification to be carried out effectively,
especially in the case of first-off inspection, pre-prepared part inspection programmes are
required.
Using and Programming CMM's.
The majority of CNC machine tools are now programmed either off-line or at the machine whilst
it is running another part program, CMM’s are only just starting to follow this path. Software for
off-line programming of CMM’s being said to lie a decade behind comparable software for CNC
machines [3]. A recent Delphi survey highlighted good off-line programming tools and the
integration of computer aided design (CAD) systems with CMM’s as major areas of
development for the 1990's [4].
CMM’s have many modes of operation, the simplest being to work interactively and use the
CMM to evaluate component features by specifying the feature type and probing the feature by
manually driving the machine. This sequence can be saved and re-run with the next component;
the so called 'teach in' mode of operation.
An inspection sequence or programme can also be modified or the CMM’s own programming
language used to write an inspection sequence from scratch.
Another mode of operation is off-line programming (OLP) which is carried out at a PC or
Workstation unconnected with the CMM. Off-line programming may be an integrated function
in a CAD/CAM system, or a separate system using geometric data transferred from a CAD
system as a basis for programming. Instead of developing probe paths with a real component
and CMM, OLP uses the CAD model. Many CAD/CAM vendors already offer off-line

27
 Paper 2: Semi-Automatic Probe Path Planning for Feature Based CMM Programming.

programming capability, vendors of robot simulation systems also offer this functionality [3, 5, 6,
7]. Commercial systems are often a combination of an established CAD system coupled with a
third party inspection program. In this way, interface problems between CAD system and off-
line programming system are avoided.
Off-line programming in a third party system, for example a CAD/CAM system, requires that
this systems native CNC programming language be converted to the CMM's language. Initially
this was done via bespoke translators, however a neutral format, the Dimensional Measuring
Interface Standard (DMIS) [8] has been defined to rationalise the communication between
CMM’s and other computer systems.
Benefits of Off-Line Programming.
Off-line programming offers many benefits including:
i Elimination of non-productive programming time on the CMM thereby offering the
potential to increase machine utilisation.
ii Avoidance of unexpected clash, reach and access problems, hence removing unproductive
re-programming time at the CMM.
iii Possibility to minimise the time required to inspect a given component, allowing greater
throughput of work at the CMM.
iv Reduction in the work involved in defining feature probing sequences, allowing better use
to be made of inspection staff's time.
The extent to which these constitute a strategic advantage will depend upon batch quantities,
component complexity and other manufacturing or production related constraints. These,
coupled with the ease of use of off-line programming tools will dictate their take-up in industry.

System Overview
A prototype off-line programming system has been developed based upon an 'open' version of a
commercial robot programming system; instead of a robot a model of the CMM, complete with
interchangeable tooling, is used. The off-line programming functionality for the CMM has been
created by the addition of two new system menus and a separate programming window.
The CMM movements required for the probing sequences generated can be simulated within the
system and corrected for access, clash and reach problems not prevented by the predictive clash
avoidance techniques described. Additional transit points and other DMIS statements required in
the final inspection program can also be added and the finished program exported in DMIS
format to the CMM.
Definition of Feature and Component Geometry.
In the off-line inspection planning system developed, component features are described in the
format given by DMIS. These feature definitions are entered via a screen form or imported from
a CAD system in the form of a DMIS text file. Component geometry is handled separately by
the host simulation system, being imported in IGES format1 [9]. From the DMIS feature
definitions the required probing points and probe paths are automatically generated for each
feature following the algorithms described.

Scope of the Algorithms.

1 Whilst this results in an almost parallel information flow, one stream to the inspection planning software and one
to the simulation system, feature recognition within the incoming IGES file was not deemed to be within the scope
of this work, having been well covered elsewhere [10].

28
 Paper 2: Semi-Automatic Probe Path Planning for Feature Based CMM Programming.

Of interest here is the automatic generation of the DMIS statements necessary to drive a CMM to
probe a given component feature. The algorithms must first place probing points on the feature,
secondly adjust these so that they can be reached with the configuration of probe used and
thirdly generate a path between these points which avoids clash with the component. The path
generation algorithms can also be used for clash evasion when moving between features if clash
occurs with a component feature that has known geometry.
Placement of probing points on a given feature.
The position of probing points on a measured feature are generated automatically with reference
to those given for a standard feature of the same type2 (Fig. 1). Any number of probing points to
be used when measuring a feature; given the minimum required to establish the size, position and
orientation of a particular feature type.
A general method to describe the location of 'n' points on a feature is not possible, so for each
value of 'n' used, a probing pattern on a standard feature must be defined. This pattern is given in
the form of a list of either a single or double co-ordinates3, which may be linear or angular
depending upon the feature type.

In order to use the above probe point location data, a feature must be of finite size. The extent of
some features can be found from their DMIS definition, whilst others DMIS considers
unbounded. DMIS allows for an unbounded feature to be limited with bounding planes. Such
boundings are used primarily for the purpose of defining tolerance zones and their use for
describing the actual feature geometry, whilst permissible, would be potentially extremely
complex.
In the system developed, the size of unbounded feature is found by prompting the user for key
dimensions when defining the feature or when invoking the 'auto-measure' function.
Alternatively, reference could be made to the actual feature geometry if this is known.4
Access to a feature with the probe used.
The basic probe point positioning algorithms assume that the probe has free access to all areas of
the feature defined. In practice, areas of a feature may be inaccessible to the probe. Several
conditions affect accessibility to a given point, and can be used to adjust the position of the
probing points so that they lie within reach of the probe.
* Actual depth of features penetrating a surface

2 In the system developed, several probing patterns can be stored and selected when generating the probing sequence.
This is a manual decision, the structure of the off-line programming system being such that the tolerance being
evaluated may not known until the user chooses to evaluate an already measured feature. Because of this, it is
impossible to use pattern/tolerance selecting routines developed by other research groups. [9]
3 Single co-ordinates on 2D features, double co-ordinates on 3D features.
4 Interaction with the original component geometry is not currently possible. The interactive definition of features
with reference to component geometry will be implemented in the future.

29
 Paper 2: Semi-Automatic Probe Path Planning for Feature Based CMM Programming.

A probe has finite length which may be greater or less than the depth that a feature penetrates a
surface, or the height of a feature which protrudes from a surface. If the length of a feature is
greater than the declared length of the probe, the probing points must be corrected to lie within
reach of the probe. (Fig. 2a) This can be done by scaling the position of all probing points so that
they lie within reach of the probe (Fig. 2b) or by shifting those that fall outside the probe's reach
(Fig. 2c).

Other situations can occur when a feature, in effect, penetrates the component. For example, the
x,y position of a hole may be established by probing a circle some distance down the hole being
measured. In this case, if the circle lies outside the reach of the probe, it must be redefined within
the probes reach or an alternative probe selected.
* Probe diameter
Apart from the obvious problem of probe diameter relative to a limiting dimension of a
penetrating feature, the diameter of the probe tip can also affect the area of a sphere or 'inverted'
conical feature that may be probed. In the case of a sphere, slightly less than 50% of its surface is
hidden from view, an 'inverted' conical feature has almost its entire probed surface hidden from
view5. In these situations, a larger probe tip diameter allows access to a proportion of these
hidden areas (Fig. 3).

* Relative orientation of probe and feature.

If the axis of the probe does not align with the feature axis, or with the normal to the plane on
which the features lie, a situation similar to that encountered with spheres and inverted conical
features occurs. The area of the feature that becomes inaccessible depends upon the relative
miss-alignment, the size of the feature and the diameter of the probe used. (Fig. 4)

5 Assuming that the feature axis aligns with the declared sensor axis, the actual area hidden depends upon the probe
tip diameter and stem diameter.

30
 Paper 2: Semi-Automatic Probe Path Planning for Feature Based CMM Programming.

Development of a clash free probe path.

Once the probing points on the feature surface have been created, the probe path between these
points can be generated. This involves placing a start and end point for the entire feature probing
sequence plus the creation of approach and retract points for each probe point. The approach
point should ensure that a normal approach to the feature surface is carried out. Finally, transit
points which avoid the feature surface may have to be added between each approach / probe /
retract cluster (Fig. 5).

* Start and End Positions

Placing sensible start and end points with respect to a feature can make the task of generating an
overall clash free path when the probing sequences for all features are executed sequentially. An
optimal location for start and end position is assumed to lie on some clearance plane lying above
the feature.
* Normal Approach Paths
In order that probe compensation be correctly applied, the probe should approach normally to
the probe surface. For each contact point, corresponding 'approach / retract' points which results
in a normal approach to the component surface must be generated.
* Clearance Transit Points
When probing the outside of objects, it is often necessary to traverse around them to avoid a
clash situation; for example measuring two points on opposite sides of a cylindrical boss. The
number and location of transit points depends upon component dimensions, probe tip diameter
and required clearance distance6.

6 The algorithms do not attempt to minimise the distance traversed by the probe by altering the sequencing in which
the probing points are visited. However, this technique has been used by other researchers.[11]

31
 Paper 2: Semi-Automatic Probe Path Planning for Feature Based CMM Programming.

Clash evasion when moving between features.

* Contact / Clash Points
Clash situations can occur when moving between one feature measurement and another. If a
clash situation is detected 7 between the probe and a component feature that is of known
geometry, the techniques described above can be applied to carry out clash evasion (Fig. 6). The
straight line transit path between two features will pass through two points on the feature surface
with which clash occurs. These two points can be taken as probing contact points when applying
the evasion techniques described above.

* Start and End Positions of Evasive Paths

The start and end positions of the evasive path should be the clearance distance away from the
impacted feature surface. As it is not known if the impact is normal to the surface, account must
be taken of feature and probe dimensions in order to find start and end points on the original
transit path that is sufficiently far away from the feature surface.
* Clearance Transit Points During Evasion
Clearance transit points can be generated in the same way as before with the number of transit
points being dependant upon component dimensions, probe tip diameter and required clearance
distance.

Conclusions.
The algorithms described above take a 'feature' view of the problem of the placement of probing
points on a feature and the probe paths between these points. These can be used as a sound
basis for the rapid generation of feature probing sequences in both off-line and on-line
programming systems. The algorithms described are especially applicable in feature based
systems or where feature recognition is implemented although the ideas expressed should also
have a place in non feature based systems.

Acknowledgements
Special thanks must be given to the authors of the GRASP robotics simulation package, BYG
Systems Ltd. of Nottingham, England, for the help and support that they have given during the
prototyping and implementation of several of the algorithms described here. Thanks also to Per
Åhman, who provided many useful pointers and carried out much of the implementation of the
ideas described above; a programmer doing a programmers job.
Acknowledgement is also duly given to my co-workers in the Department of Manufacturing
Engineering at Luleå University of Technology and especially to NUTEK for funding this work.

7 Clash detection routines exist in the robot simulation system used.

32
 Paper 2: Semi-Automatic Probe Path Planning for Feature Based CMM Programming.

References
1. J. Bosch, "The Case for CMM’s." Tooling and Production October 1988.
2. W. Tandler, "Critical Concepts of High Performance Co-ordinate Measuring Systems."
Multi-Metrics Incorporated, Redwood City, California, USA. 1984.
3. F. Mason, "Program Your CMM Off-Line." American Machinist October 1992.
4. 1991 Report by Future Technology Surveys Incorporated, Lilburn, GA, USA.
5. J. Granquist, "Closing the Loop of Mold Design, Manufacturing and Inspection."
American Machinist April 1993.
6. "Fine-Tuning Quality." Manufacturing Engineering November 1992.
7. "CMM Support Systems Speed Throughput." Manufacturing Engineering March 1992.
8. "The Dimensional Measuring Interface Specification." ANSI CAM-I 101 1990
9. I. Walker & A.F. Wallis, "Applications of Solid Modelling to Co-ordinate Measuring
Inspection." International Journal of Machine Tools & Manufacturing Vol. 32 No 1/2.
10 K.E. Hummel & C.W. Brown "The Role of Features in the Implementation of
Concurrent Product and Process Design." Proceedings from Symposium on Concurrent
Product and Process Design, ASME Winter Meeting 1989
11 F.L. Merat & G.M. Radack "Automated Inspection Planning within the Rapid Design
System." IEE CH 3051-0/91/0000-0042

33
 Paper 3

Integration of Design, Inspection and

Quality Management Systems

34
 Paper 3: Integration of Design, Inspection and Quality Management

Integration of Design, Inspection and

Quality Management Systems.
Eur. Ing. David Ian Legge
Department of Manufacturing Engineering,
Luleå University of Technology, Sweden

Abstract
This paper discusses the standard for exchange of product model data ( STEP - ISO 10303 ) and
the proposed STEP application protocol (AP) for inspection process planning (AP219). The need
for a concerted industry lead initiative to develop this AP is suggested. An industry consortium
should include users and vendors of the hardware and software associated with inspection related
activities and would build upon work already ongoing including work related to STEP. The goal
of such a consortium would be to develop and promote the use of STEP for inspection process
planning and related activities and will act as a focus for development of both AP219 and
software tools to support it's adoption and implementation.

Keywords
Co-ordinate Measuring Machines, Off-Line Programming, STEP

The Scope and Nature of Inspection

Quality is no longer 'inspected in' to a product at the end of production. However, inspection is
still a very important activity and increasingly sophisticated automated and semi automated
inspection systems have become available. This equipment may use contact or non-contact
methods, may be dedicated to a particular purpose or may be programmable and may stand
alone or be integrated into other pieces of equipment.
One of the most common pieces of programmable inspection equipment in use today is the co-
ordinate measuring machine or CMM. Whilst this paper discusses the integration of design,
inspection and quality management systems using STEP with particular reference to CMM's, the
discussion is equally applicable to other programmable inspection equipment.

Co-ordinate Measurement - Current Status

CMM’s are commonly used for component inspection both in metrology rooms and,
increasingly, on the shop floor as independent resources or as part of closely coupled
manufacturing lines and cells. Inspection using a CMM is typically a three stage activity;
programming, execution of the programme and evaluation of the results. These activities are
often carried out on separate systems which leads to complex interface problems. (see Figure 1.)
Programming of CMM’s is, even today, often carried out in a 'teach-in' mode. This is both time
consuming and, for small batch quantities, results in a high proportion of unproductive time
whilst an inspection strategy is planned and a valid programme generated. Manual editing the
inspection program, which is invariably in the CMM’s own native language, is often at a similarly
basic level, often relying upon relatively unfriendly text editors such as 'vi.'
Software is now becoming available which allows 'off-line' development of part programmes for
CMM’s in a similar manner to the off-line part programming of CNC machine tools
(CAD/CAM). Such off-line programming (OLP) tools are becoming more commonplace,

35
 Paper 3: Integration of Design, Inspection and Quality Management

although their use is still far from widespread. The leading OLP tools offer a 3D graphical
presentation of the component being inspected as well as varying levels of automation for
creating inspection paths for individual features and of overall inspection planning.
Like CAD/CAM, OLP systems fall into two classes; those that are integrated into a host CAD
system, via functionality written into the CAD system or in the form of a third party system
which is closely integrated through bespoke data transfer routines or through access to common
database, and those which are 'stand-alone' but which can import data from the source CAD
system in a neutral format such as the initial graphics interface standard (IGES).
Commercial OLP systems usually offer semi automation of the planning process. In the case of
independent, 'stand alone,' systems, the user must typically decide upon the overall sequence of
inspection as well as identifying which features are to be measured and how they are to be
evaluated. The lack of tolerance data in geometry transferred from CAD in IGES format means
that tolerances must be entered manually. Research systems have demonstrated high levels of
automation when a fully toleranced geometric model is available. This functionality is likely to
become more commonplace in commercial systems in the future.
OLP systems generate inspection process plans which have then to be passed onto the CMM.
This is done by producing output in either the CMM's native programming language or in the
dimensional measurement interface standard (DMIS - ANSI 101 1990). DMIS programmes are
invariably translated into the CMM’s native language in a pre-processor before being run.
Most companies will claim to have unique inspection problems, however, virtually all inspection
situations fall into one of two categories:
a) Process evaluation / validation; where the principle aim is the comparison of production
process and product. This would include development of forming tools and first off-
inspection of products manufactured on CNC machine tools.
b) Product validation; where inspection is carried out for acceptance / rejection and product
tracability or statistical process control (SPC) of production processes of known capability.
Evaluation of a component is typically by assessment of individual features against tolerance,
although more complex evaluation may include fitting of several inspected features to nominal
geometry simultaneously. Evaluation may be carried out by the CMM control software or by
analysis of raw or part processed data output from the CMM. This output is typically in a user
defined format, although the DMIS covers output of raw or evaluated data in a comparable
format to the inspection plan used to generate the data. Feedback of data to computer aided
design (CAD) systems is more typically the case in the former category of problem.

36
 Paper 3: Integration of Design, Inspection and Quality Management

CAD CMM CAD OLP CMM Evaluation

Validation
OLP Evaluation
Presentation
Validation
Presentation
IGES CMM Native Bespoke
DMIS

CAD OLP CMM

CMM
CAD OLP
Evaluation
IGES DMIS Validation
Presentation
Evaluation Bespoke DMIS
Validation
Presentation CMM Native

Figure 1. Possible integration combinations for design, programming and inspection systems.

The Standard for the Exchange of Product Model Data

STEP (ISO 10303) aims to '' specify a form for the representation and unambiguous exchange of
computer-interpretable product information throughout the life of a product. The form is
independent of any particular computer system. This form enables consistent implementations
across multiple applications and systems .... The approach for representation is to provide one
definition of product data common to many applications. ''
From this it follows that STEP could (and should) be used in place of the many existing standard
and non standard formats in use for the integration of design and inspection related systems
today. (see Figure 2.) This consistent format for all information involved in the inspection
process can only offer benefits in the long run.
3D CAD to OLP - AP203.
STEP is already well developed in the area of definition of nominal geometry. Vendors of off-
line programming systems are already working towards STEP interfaces from the AP's in the
initial release of STEP. At this level, STEP would appear to offer no immediate benefits over
IGES, the data format is not dissimilar and only nominal geometry is covered.
However, the adoption of STEP as a commercial necessity is encouraging. This also will allow
experience to be gained in the techniques and data structures typical of STEP.
Toleranced 3D CAD to OLP - AP 2??
Leading commercial CAD systems can now offer tolerancing of 3D geometric models,
functionality which is of great help in the automation of down stream planning activities. Given a
toleranced CAD model, it is possible to achieve high levels of automation in OLP systems.
Research systems have demonstrated automatic selection of the features to be inspected,
selecting the number of inspection points necessary and determining an overall sequence for the
inspection task. This level of functionality should become commonplace in commercial OLP
systems the future.
Integrated OLP systems will be the first to benefit from the availability of 3D toleranced models
and subsequently, as the ability to transfer toleranced geometric models from CAD to down
stream activities becomes addressed by STEP AP's, non-integrated off-line programming systems

37
 Paper 3: Integration of Design, Inspection and Quality Management

will be able to use this information, allowing the automation of overall inspection strategy and
development of individual probing sequences.
OLP to CMM - AP219
The STEP application protocol, AP219, for inspection process plans has already been identified,
although little work has been done towards its development. The current standard for transfer of
inspection process plans, DMIS, will obviously act as a basis for AP219, however, DMIS is
recognised as lacking explicit structure and being underdefined in some areas whilst being
overdefined in others. Users experiences of DMIS as well as developments in off-line
programming systems and CMM control software must also be taken into account.
CMM and Evaluation - AP219
AP219 should also cover the results of the inspection, as either raw data, semi-evaluated data or
fully evaluated data. This information should obviously be suitable for evaluation in external
systems as well as being fed back to CAD. Feed-forward to product management systems,
(MRP and quality control software.) is also a possibility.
Inter-operability with other AP's
Inspection process planning is not only carried out on stand alone CMM's. In process gauging is
becoming increasingly important in both machines and cells and AP219 should also be usable in
this type of system. It would seem likely that this level of development will take place once the
basic format for AP219 has been established, although this base must take some account of the
other AP's which concern process planning.
CAD CMM CAD OLP CMM Evaluation
Validation
OLP Evaluation
Presentation
Validation
Presentation
AP2?? AP219 AP219
AP219 AP219

CMM
CAD OLP
Shop floor control Evaluation
Validation
Presentation
AP203 AP219
CAD OLP CMM

AP203 AP219

Evaluation
Validation
Presentation AP219 Quality audit / tracability
|

Figure 2. Possible integration combinations using STEP.

38
 Paper 3: Integration of Design, Inspection and Quality Management

Proposed Industry Consortium

Formal development of the AP for inspection process planning, AP219, is currently in a state of
abeyance and is regarded as being of low priority by WG3. Work related to STEP which
includes reference to inspection planning is being carried out by an industry consortium centred
on CAM-I ( The organisation responsible for the development of DMIS. ) in the USA. The
european ESPRIT Vision Based On-Line Inspection of Manufactured Parts (VIMP) project has
apparently included some work although this is not ongoing.
The scope of AP219, and the relatively limited number of hardware (CMM) and software (OLP)
vendors potentially involved in the development of this AP, make it an ideal candidate for the
rapid and successful development and demonstration of STEP based data transfer.
It is proposed that an industry consortium be formed to adopt, develop and promote the work
which has been carried out to date on this AP in Europe and the USA. The consortium would act
as a catalyst for the implementation of AP219 in commercial systems and in company databases
through the development of the AP and of software tools to assist in the implementation and use
of AP219.
The automotive industry is ideally placed to provide the impetus necessary for the proposed
consortium to be effective. Not only is the automotive industry well placed to motivate
commercial development of STEP AP219, it also has representative hardware and software
systems from the majority of the off-line programming and CMM suppliers.
Industry lead initiatives are not unique. The US Navy Industry Digital Data Exchange Standards
Committee (NIDDESC) is looking at the use of STEP for ship outfitting, whilst the UK process
industries are promoting the use of STEP through the PISA programme.

Development of AP219
On the assumption that sufficient interest to develop AP219 exists, the first step must be to
establish the current status of ongoing work related to STEP and inspection planning
A number of groups working in this area are known, whilst the remainder should be contactable
without too much difficulty. Once the status of the work carried out to date is known, a more
detailed strategy for development can be established. Input from individuals or groups with
experience of development of STEP AP's would obviously be actively sought.
It would seem likely that some software tools specific to AP219 could beneficially be developed.
These would allow parsing of STEP files, interface to commonly used databases, input,
interrogation and editing of data within STEP format files or databases. Such tools would be of
use to both system developers and users.
The original estimate for development of AP219 was in the order of three years. This still
remains a realistic, but not overly conservative, estimate.

Conclusions
Development of STEP in the areas related to inspection, e.g. transfer of toleranced 3D geometry
will lead to a significant increase in the functionality and level of automation offered by off-line
programming systems including off-line programming of co-ordinate measuring machines. For
this to become a reality, STEP must become more widely adopted and seen as a necessary
development by industry.

39
 Paper 3: Integration of Design, Inspection and Quality Management

The STEP application protocol for inspection process planning, AP219, is not being actively
developed at present. Development of AP219 will provide an ideal opportunity for individuals
and companies to develop knowledge of and expertise in the use of STEP.
The scope of AP219, and the relatively limited number of hardware (CMM) and software (OLP)
vendors potentially involved in the development of this AP, make it an ideal candidate for the
rapid and successful development and demonstration of STEP based data transfer.
Development of AP219 will ideally be with the support of an industry consortium.
The automotive industry is ideally placed to provide the impetus necessary for the proposed
consortium to be effective. Not only is the automotive industry well placed to motivate
commercial development of STEP AP219, it also has representative hardware and software
systems from the majority of the off-line programming and CMM suppliers.
A range of either public domain or consortia specific software tools will be developed as a means
of providing support to both system developers and users

40
 Epilogue:
” In the search for truth there are certain questions that are not important.
Of what material is the universe constructed ? Is the universe eternal ? Are
there limits or not to the universe ? What is the ideal form of organisation
for human society ? If a man were to postpone his search and practice for
Enlightenment until such questions were solved, he would die before he
found the path. ”

- Buddha

41