Advanced STEP parameterised and constrained features for reverse engineering

Int. J. Computer Applications in Technology, Vol. x, No. x, 200x 1

Copyright © 200x Inderscience Enterprises Ltd.

Advanced STEP parameterised and constrained features for Reverse Engineering

Sébastien Remy and Guillaume Ducellier Charles Delaunay Institute, Troyes University of Technology, LASMIS, FRE CNRS 2848, 12 rue Marie Curie, BP 2060, F.10010 Troyes Cedex, France Fax: (+33)-3-25-71-56-75 E-mail: [email protected] E-mail: [email protected]

Sébastien Charles IUT of Mantes en Yvelines, Laboratory of Systems Engineering of Versailles, University of Versailles Saint Quentin en Yvelines, F.78200 Mantes la Jolie, France Fax: (+33)-1-30-98-13-61 E-mail: [email protected]

Benoit Eynard* Charles Delaunay Institute, Troyes University of Technology, LASMIS, FRE CNRS 2848, 12 rue Marie Curie, BP 2060, F.10010 Troyes Cedex, France Fax: (+33)-3-25-71-56-75 E-mail: [email protected] *Corresponding author

Abstract: Reverse Engineering (RE) is a research field where physical models are measured or digitised in order to be reconstructed for finally obtaining a CAD model. Regarding the large range of existing approaches, the paper deals with features known by their mathematical equation (vectorial or Cartesian). Typically, these equations are obtained when a part composed by features is measured using a touch probe integrated on a Coordinate Measuring Machine (CMM). The paper presents an interactive approach for RE of mechanical part designed using feature. It introduces a CAD model reconstruction methodology based on new STEP AP203 ED2 standard integrating parameterised features. This one enables the parameterisation of the rebuilt features which is needed for a complete integration of the new CAD model in a collaborative product development or a PLM approaches. The obtained CAD model is a 3D part fully parameterised. Last, the reconstruction methodology is detailed and is illustrated with examples.

Keywords: reverse engineering; RE; CAD; interoperability; STEP; collaborative design.

Reference to this paper should be made as follows: Remy, S., Ducellier, G., Charles, S. and Eynard, B. (xxxx) ‘Advanced STEP parameterised and constrained features for Reverse Engineering’, Int. J. Computer Applications in Technology, Vol. x, No. x, pp.xxx–xxx.

Biographical notes: Sébastien Remy obtained his PhD studies in 2004 at the University of Nancy (France), where he explored the 3D digitising field and especially the automation of the inspection of free form surfaces. Then, he spent one year at University of Windsor, Canada, as a Post Doctorate research Fellow where he worked in the Reverse Engineering (RE) domain. He currently is working at the Troyes University of Technology, France, as an Assistant Professor. He manages CAD lectures for undergraduated students and Advanced CAD application lectures for graduated students. His research field remains 3D digitising and RE.

Author: Please reduce abstract of no more than 100 words.

2 S. Remy, G. Ducellier, S. Charles and B. Eynard

Guillaume Ducellier currently is PhD student at the Troyes University of Technology, France. He obtained is MSc in Mechanical Engineering in 2004. He makes his research works under the supervision of Dr. Benoît Eynard at the Laboratory of Mechanical Systems and Concurrent Engineering (LASMIS) of the Charles Delaunay Institute. His research interests include collaborative design, product lifecycle management and knowledge management.

Sébastien Charles recently joins as Assistant Professor at the University of Versailles Saint Quentin en Yvelines, France. He is also a member of Laboratory of Systems Engineering of Versailles (LISV). He received a PhD Degree from the University of Technology of Troyes (France) in 2005. His research interests include concurrent engineering, collaborative product development, product lifecycle management, design and simulation integration, management of CAD and FEA data, and new STEP standards.

Benoît Eynard currently is an Associate Professor of Mechanical Engineering and Information Technology in the Department of Mechanical Engineering at the Troyes University of Technology, France. He is also a Member of the Laboratory of Mechanical Systems and Concurrent Engineering (LASMIS) of the Charles Delaunay Institute. He received a PhD Degree from the University of Bordeaux 1 in 1999. This thesis dealt with Engineering Design and Computer Integrated Manufacturing. Currently, his research interests include concurrent engineering, product data exchange, product lifecycle management, engineering knowledge modelling and reuse.

1 Introduction

The creation, management, improvement and exchange of CAD models within the product lifecycle became one of the major stakes for many companies and especially in mechanical engineering field. Indeed, in a context of collaborative engineering and extended enterprise, which require large range of heterogeneous CAx systems, the major stakes are to reduce transfer times and to improve quality of data exchange by using neutral file formats (Sheder, 1995; Spooner and Hardwick, 1997). The STEP standard is recognised as one of the best solution for these kinds of exchange issues. The latest developments of STEP Integrated Resources (IR) such as Part 108 provide some new prospects for data exchanges (see Section 3.2). Once issued, this Part will allow the exchange of 2D (sketches) and 3D data including parameterised geometry and constraints. The resulting 3D models will be non-frozen and directly re-useable in any CAx systems. In a RE approach, creation of CAD models from real parts is often difficult because of the lack within methods and tools to recover geometries and parameters. Currently, geometric models rebuilt based on RE approach are generally frozen (parametric surfaces approaches) or not re-usable (meshed surfaces approaches). This is why, to fill the gaps of this field, we propose methods and tools to interactively create non-frozen and neutral CAD models from real parts in this paper.

This paper is organised as follows. Section 2 presents new needs regarding advanced CAD data exchanges. Afterwards, Section 3 introduces the STEP standard and describes some of new Parts currently developed, and especially the Part 108, main purpose of this study. Section 4 describes how these introduced STEP Parts fulfilled the new CAD exchanges needs. Section 5 details the methodology and technical approach used to recover parameters and geometric features from real parts. Section 6 proposes the software application implementing the

presented approach. Finally, Section 7 concludes the paper and introduces future developments.

2 Reverse Engineering (RE) application

RE refers to create a CAD model from an existing physical object, which can be used as a design tool for producing a duplicate of an object, extracting the design feature of an existing model, or re-engineering an existing part. In other words, RE takes information from the real world like a point cloud of an object’s surface captured using a 3D digitising technology as an input and creates a geometric model, which should be compliance with the requirements for a rapid prototyping system or CAM. Since the cloud data are generally dense and unorganised, reconstructing a geometry model for efficient and accurate prototype manufacturing becomes a major research issue.

2.1 Main RE approaches

In general, approaches for modelling an object from a cloud data can be classified in two categories, i.e.,

• surface reconstruction based on an implicit function (e.g., parametric function) (Sakar and Menq, 1991) or

• surface modelling employing a polyhedral mesh (e.g., triangular mesh) (Urk and Levoy, 1994).

The segment-and-fit approach described by Hoffman and Jain (1987) is widely used in the former method. Typically, the cloud data is segmented into several patches bounded by clearly defined curves, each representing a discrete surface region present the physical object. Modelling methods, such as those employing parametric (Varady et al., 1997) or quadric (Chivate and Jablokow, 1993; Weir et al., 1996) functions are then applied to fit surfaces to the patches using NURBS (Piegl and Tiller, 1995). This approach could be a difficult and tedious task. It is noticed that some commercial

Advanced STEP parameterised and constrained features for Reverse Engineering 3

RE packages combine the polyhedral mesh and parametric surface reconstruction (http://www.paraform.com).

For the approach employing polyhedral meshes, the inherent data structure produced by the vision system plays a critical role on the meshing techniques. The structure in the data can range from being highly organised, such as an array of points, to little structure, such as cloud data. For a highly structured data set, such as a range image composed of a regular grid of data points, a polygonal model can be created in a straightforward manner by linking data points in a neighbourhood to form the mesh. If an object is digitised through the acquisition of multiple range images, then an appropriate registration and alignment technique must be implemented to merge the set of adjoining polygonal domains (Urk and Levoy, 1994; Soucy and Laurendeau, 1995).

Algorithms that have been developed for modelling less structured three-dimensional data sets assume that no a priori information regarding the connectivity of points in the data set is available. The only assumption is that there exists a sufficiently high data sampling resolution to permit unambiguous reconstruction of the model. For example, Fang and Piegl (1995) extended 2D Delaunay triangulation algorithm to three-dimensional data. Cignoni et al. (1998) described another Delaunay triangulation technique based on a divide-and-conquer paradigm. Lawson (1977) used geometric reasoning to construct a triangular facet mesh, and subsequently, Choi et al. (1988) extended the same method using a vector angle order, instead of Euclidean distance, to determine the linkage of data points. Hope et al. (1992) developed a signed distance function by estimating the local tangential plane and using a marching cube method to extract a triangular polyhedral mesh.

2.2 Proposal for RE application based on parameterised feature

All these references are generally old and there are many recent papers that deal with RE application. These papers focus on the improvement of the data like the quality of the meshed point cloud (Lai and Lai, 2006) or the surface fitting operation (Galetto and Vezzetti, 2006, for example). The two main approaches described in this section are good tools for copying a given object but they present drawbacks for real RE. The parametric approach provides accurate surfaces but it is long and parametric surfaces are sometimes not enough to totally describe a feature from an object with a designer point of view. For example, a cylinder rebuilt using NURBS surfaces can not provide a diameter information. The mesh approach is very quick and efficient to copy an object but in fact, the geometric model obtained is only a set of triangles that is very poor in terms of information regarding to design or manufacturing features. An issue remains regarding to the possibility to reuse the obtained CAD model in a Re-Engineering operation. In fact, a meshed model and a parametric surfaces based model do not provide the diameter of a hole (a set of triangles or a parametric surface are not a cylinder). These models do not provide the possibility to put

a parallelism constraint between two planes because a set of triangles and a parametric surface are not planes. In the following section, the neutral formats and especially the STEP file format are studied to show their interest regarding this issue.

3 The needs of CAD field

3.1 Neutral formats

Regarding an industrial context, numerous partners are often involved in projects. The heterogeneity of their software application means a significant diversity of native format. This diversity implies numerous translators to improve the exchanges of product data. In order to reduce the number of required translators, standardisation of file formats is necessary. Indeed, when N different software are involved for carrying out a product development, Nx(N – 1) translators are required without using a common format. With a neutral format, only N translators are necessary. Figure 1 illustrates this issue.

Figure 1 Benefits of neutral format

The new needs regarding advanced CAD data exchanges evolved these last years. Indeed, in the past, the geometric data transfers based on 3D neutral format were limited to the exchange of frozen models with no possibilities to modify dimensions or to change parameters of features (such as the depth and diameter of a hole).

3.2 Construction history

Current 3D CAD models consist in a set of geometric and topological entities, called ‘form features’, organised in a hierarchical way through a model tree. This tree structure gives a history of the entities created along the modelling of the 3D part. Parent-children relationships are used to establish links between the components of the model (nodes), in order to understand the history of model creation.

3.3 Parameters, constraints and relationships

Regarding the challenges of collaborative design, the designers require the definition of explicit parameters, constraints and relationships between geometric and dimensional features into a 3D model. The ‘explicit’ term

4 S. Remy, G. Ducellier, S. Charles and B. Eynard

refers to parameters, constraints and relationships specified by the user of the CAD software. On the opposite, ‘implicit’ term refers to an automatic creation implied by geometric constructions. The explicit elements are generally created to guarantee that the modelled parts will ensure the functional specifications of product. Thus, constrained models can be exchanged without risk between the various designers and project teams.

3.4 Sketch data

The construction of 3D shapes into a CAD models is most of the time based on 2D sketches. These sketches consist in a set of geometric features, dimensions, constraints and relationships with a history of construction. It is therefore interesting to exchange both the geometry of the 3D models and the sketches defining the features. Moreover, the way the sketches are defined often reflects the design intents. Unfortunately, no standard format currently takes into account this kind of data. However, the STEP standard is on the right way to provide a solution to this issue.

4 STEP survey

4.1 The STEP standard

The STandard for the Exchange of Product data model (STEP) is an international standard of International Standard Organisation (ISO) referenced as ISO 10303. The STEP standard aims to define a non-ambiguous, computer-interpretable representation of the data related to the product throughout its lifecycle (Fowler, 1995). STEP allows the implementation of consistent information systems through multiple applications and materials. This standard also proposes various means for the storage, exchange and archiving of product data in a strategy of long-term re-use.

The STEP Application Protocols (AP) specifies data-models (EXPRESS Schema) applicable to an industrial field of activity and product lifecycle phase (Hardwick et al., 2000). The AP are formal representations of the data of a particular field which use and add semantics to data entities defined in the STEP IR. These IR are shared models which have the role of ensuring interoperability between the various AP (Zhang et al., 1998). To summary, the APs are used as bases to implement the STEP import and export modules in software (Yeh and You, 2000).

4.2 The latest improvements of the STEP standard

This section is dedicated to the presentation of the STEP Parts currently developed. These Parts will propose a solution to a part of the needs of parameterised geometric data exchange presented in Section 3.

4.2.1 Exchange of non-frozen models: STEP Part 108

The STEP Part 108 is entitled: “Parameterisation and constraints for explicit geometric product models”

(Pratt, 2004) and ISO 10303-108. This part belongs to Applicative IR in the field of the representation and exchange of 3D product data. The STEP Part 108 provides general representations for parameterised values and for constraint relationships between features in 3D models and in 2D sketches. It is important to transfer these parameters and constraints because they enable to understand the designer intents. These intents, created in the design software, must appear in all the other applications of the product lifecycle. The STEP Part 108 reached the International Standard status – IS under ISO 10303-108 (2004).

The STEP Part 108 deals with:

• the parameters and the relationships defined between sketch, part and assembly

• the explicit constraints defined between geometrical features of sketch, part or assembly

• the parameterised and constrained sketch features used as reference elements in the standard 3D CAD features (like protrusions for example).

The STEP Part 108 provides the necessary bases for the development of an Application Protocol dedicated to the geometric data exchanges, which can take into account parameter, relationship and constraint data. It is partly defined to fill the gaps of the original version of the STEP AP203 Application Protocol mainly dedicated to the geometric data exchanges of 3D CAD models without sketch and constraint. Indeed, the AP203 allows only the exchange of frozen geometric models for which it is not possible to modify the dimensional relations and the constraints. Therefore, the re-use of exchanged models is grandly limited. Part 108 could be addressed by the new version of STEP AP203 Edition 2 (or in a further Edition) as an Integrated Resource. This new version of the Application Protocol should take into account the constraint, parameter, relation, construction history and sketch data for 3D models exchanges.

4.2.2 Additional STEP new parts

Other new Parts are currently developed to complete the gaps of the AP 203 (Pratt et al., 2005). These STEP Parts combined with the Part 108 will improve the quality of the neutral CAD exchange. These are described below:

• The Part 55 – Procedural and Hybrid Representation (International Standard (IS)). Defines fundamental resources for the representation of models in terms of the sequences of constructional operations used to construct them. Entities defined in other parts of ISO 10303 are used to represent the constructional operations themselves.

• The Part 109 – Kinematics and geometric constraints for assembly models (IS). Specifies the resource constructs for the representation of detailed geometric relation between constituents of an assembly model including geometric constraints between constituents.

Advanced STEP parameterised and constrained features for Reverse Engineering 5

• The Part 111 – Elements for the procedural modelling of solid shapes (Draft International Standard (DIS)). This part of ISO 10303 specifies resource constructs for representing the complex shape elements, sometimes known as form features, which are supported by the solid modelling capabilities of modern CAD systems. The elements are defined in such a way as to facilitate the exchange of solid models of products represented in terms of their constructional history. This part completes Part 42 (Geometric and topological representation) which specifies primitive solid geometry elements (like cylinders, cones, spheres, etc.) in the field of feature based CAD models.

• The Part 112 – Modelling commands for the exchange of procedurally represented 2D CAD models (DIS). Specifies the resource constructs for the representation of 2D modelling commands for use in the exchange of procedurally represented 2D CAD models. A procedural model is defined in terms of the sequence of 2D modelling operations used to build it.

All the IR described in this section will contribute to CAD data exchange improvements by using non-frozen models. These IRs would be referred by the new version of STEP AP203 Edition 2.

5 Implementation test of the Part 108 capabilities for non-frozen models

Regarding the benefits of the future implementation of AP203 Edition 2, implementation tests have been carried out to evaluate the compatibilities offered by STEP Part 108 in a RE application. This section details the tests carried out in order to validate the last STEP developments regarding standardisation of non-frozen model exchanges in a RE approach.

5.1 Presentation of the implementation case-test

The main goal of the implementation described in this paper is to evaluate the capabilities offered by STEP Part 108 to create non-frozen models based on digitising of real parts. To assess these capabilities, methodology was defined and tools developed. The first step of the method consists in recovering geometry, topology, constraints and dimensional parameters from measures of real parts. Once recovered, these data are interpreted by CAD system through a translator which converts unprocessed data to native format. The geometries recovered are only surfaces (plane, cylinders, etc.), the CAD user must perform Boolean operations on them to create the shell of the part. Once the shell is completed, a solid is created. The designer can enrich the model by adding parameters, constraints and relationships. In that way, the CAD user can add his or her design intent and create a parameterised and constrained features which can be modified and reused in another context. The last step of the approach consists in translating these data to an enhanced STEP AP203 file (completed with

the specifications of Part 108 and some additional specifications of Parts 55 and 42). New Edition of AP203 is not currently achieved, then it has been necessary to implement part 108 in an advanced STEP AP203 file format Thus, the capabilities of Part 108 can be evaluated through the translation process by checking the presence and reliability of required entities.

The implementation process has been carried out as follows:

• choice of a relevant mechanical part representative of geometries, parameters and constraints data defined in the standard

• measure of the part to extract geometries, parameters and constraints in an ASCII file

• implementation of a translator to get back these data and create a CAD model; Creation of a mapping table between unprocessed ASCII and CAD native formats

• checking of the consistency of the model by doing comparisons with the real part

• enhancement of the CAD model by adding new parameters, constraints and relations

• implementation of a translator to get back these data and create an advanced STEP AP203 file; Study of the STEP Part 108 in order to identify the data needed by the use case implementation; Creation of a CAD native to STEP formats mapping table

• creation of the file

• identification of lacks of STEP part 108 regarding the test requirements.

5.2 New STEP entities implemented in the case-test

The STEP Part 108 entities required by the demonstrator implementation are of three different kinds: constraints, parameters and relationships. The static organisation of STEP entities described bellow is briefly illustrated on Figure 4.

The PARAMETERIZATION_SCHEMA defines two types of constraints: explicit and implicit. As said before, an implicit constraint is built into the process of a constructional operation. An explicit constraint is one that is deliberately added by the system user. In the demonstrator context, only explicit constraints are taken into account. The supertype entity of this kind of constraint is EXPLICIT_GEOMETRIC_CONSTRAINT. All subtypes of EXPLICIT_GEOMETRIC_CONSTRAINT share two attributes, one specifying a set of one or more constrained elements and the other a set of zero or more reference elements. If reference elements are present in the instance of such a constraint, then the constraint is said to be directed; if not, it is said to be undirected. In the case-test, only directed form of the constraints is used.

The main constraints entities used in the demonstrator are the following (each attribute of these entities are described):

6 S. Remy, G. Ducellier, S. Charles and B. Eynard

• PARRALLEL_OFFSET_GEOMETRIC _CONSTRAINT_WITH_DIMENSION. This entity constrains a set of curves or a set of surfaces to be parallel offsets of each other and specify the value of the offset. Attributes of this entity are: PARRALLEL_OFFSET_GEOMETRIC _CONSTRAINT_WITH_DIMENSION(‘name’, ‘description’, (constrained elements: #CURVE or #SURFACE), (reference: #CURVE or #SURFACE), offset type: CURVE_2D_OFFSET or CURVE_3D_OFFSET or SURFACE_OFFSET, attribute: OFFSET_VALUE, offset direction constrained: .T. if a direction is required and .F. if not). OFFSET_VALUE is subtype of POSTITIVE _LENGTH_MEASURE, it means that the value must be positive. Some entities are optional like the description or the reference entity in the undirected case.

• COAXIAL_GEOMETRY_CONSTRAINT. This entity constrains a set of AXIAL_GEOMETRY _CONSTRAINT_ELEMENT instances to share the same axis. The AXIAL_GEOMETRY _CONSTRAINT_ELEMENT entity addresses all kind of geometries (point, line, circle, etc.) and primitive entities (cylindrical surface, torus, right circular cone, etc.) which can be associated with an axis (not curves for example). Points are constrained to lie on the axis, lines to be coincident with it and planes perpendicular to it. For a circular element, the axis is taken to be the line through its centre and perpendicular to its plane of definition: constrained circles are not required to be coplanar with each other. Any line through the centre of a sphere or spherical surface may be regarded as its axis. Attributes of this entity are the following: COAXIAL_GEOMETRY_CONSTRAINT(‘name’, description’, (constrained elements: #AXIAL _GEOMETRY_CONSTRAINT_ELEMENT), (reference: #AXIAL_GEOMETRY_CONSTRAINT _ELEMENT). Reference element is optional in the undirected case.

The definition of parameters and their association with quantities of the model are specified by the PARAMETERIZATION_SCHEMA. This schema specifies how to represent a parameter through MODEL _PARAMETER entities. An instance of a subtype of MODEL_PARAMETER is regarded as a mathematical variable; it has a domain of validity and a current value. Two instantiable subtypes are defined, BOUND_MODEL_PARAMETER and UNBOUND _MODEL_PARAMETER. An instance of BOUND _MODEL_PARAMETER is associated with an attribute of a step entity of the model (like the radius of a circle for example). An instance of UNBOUND_MODEL _PARAMETER, as its name implies, is not bound to an entity instance attribute. Instead, it participates in a

mathematical relationship with other parameters, some of which may be instances of BOUND_MODEL _PARAMETER. For example the unbound parameter ‘t’ can be used to create a relation between the coordinates of a 2D point: x = 3t – 2, y = 2t. The developed demonstrator only addresses bound model parameters. STEP entities required to define a bound model parameter are the following (attributes are describes):

• BOUND_MODEL_PARAMETER. This entity allows the creation of a parameter. The Attributes of this entity are the followings: BOUND_MODEL_PARAMETER (‘name’, domain of validity: #FINITE_REAL _INTERVAL, ‘description’, (DER)PARAMETER_ CURRENT_VALUE); The parameter current value is for example the current value of the radius of a circle. It is a derived attribute; it means that it can be recovered from another entity.

• BOUND_PARAMETER_ENVIRONMENT. This entity is used to associate a parameter with a variable attribute of a geometric entity. The attributes of this entity are: BOUND_PARAMETER _ENVIRONMENT(#BOUND_MODEL _PARAMETER, #INSTANCE_ATTRIBUTE _REFERENCE).

• INSTANCE_ATTRIBUTE_REFERENCE. This entity specifies which attribute of a geometric entity is linked to a parameter. The attributes of this entity are the followings: INSTANCE_ATTRIBUTE_ REFERENCE (‘GEOMETRIC_MODEL_SCHEMA. GEOMETRIC_REPRESENTATION_ITEM. ATTRIBUTE’, #GEOMETRIC_REPRESENTATION _ITEM). A geometric representation item is the supertype of all the type of geometric and topologic representations defined by STEP.

The parameter relationships defined in PARAMETERIZATION_SCHEMA are mathematics expressions involving bound or unbound model parameters. The mathematics operands are defined by the ISO 13584-20. The main relationship used in the test case is the following:

• EQUAL_PARAMETER_CONSTRAINT. This entity constrains two parameters to be equal. The attributes of this entity are: EQUAL_PARAMETER _CONSTRAINT (‘name’, ‘description’, (constrained entities: MODEL_PARAMETER), (reference: MODEL_PARAMETER)).

In addition to Part 108, the case-test also implements Part 55 capabilities to represent the sequence of Boolean operations performed on plane and cylindrical surfaces. Boolean operations are not taken into account in the first version of STEP AP203 although they are defined in Part 42. The Parts 55 and 42 entities implemented in the advanced STEP file are the followings:

Advanced STEP parameterised and constrained features for Reverse Engineering 7

• BOOLEAN_RESULT. This entity specifies Boolean operations performed between geometric elements of a model. The attributes of this instance are: BOOLEAN_RESULT(‘name’, type: .UNION. or .INTERSECTION. or .DIFFERENCE., boolean operand: #GEOMETRIC_REPRESENTATION _ITEM, boolean operand: #GEOMETRIC _REPRESENTATION_ITEM). The Boolean entities which can be addressed by the BOOLEAN_RESULT are the following: SOLID_MODEL, HALF_SPACE_SOLID, CSG_PRIMITIVE, BOOLEAN_RESULT or HALF_SPACE_2D. CSG_PRIMITIVE entities are Constructive Solid Geometry models which allow the creation of primitive volumes like cylinders, cones or spheres. These entities defined in Part 42 are not addressed by the preliminary edition of STEP AP203 but will be addressed by the fore coming edition to fulfill the current CAD system needs.

• PROCEDURAL_SOLID_REPRESENTATION _SEQUENCE. This entity allow the representation of a sequence of operation to specify in which order they must be performed to guarantee consistency of the model. The attributes of this entity are the followings: PROCEDURAL_SOLID_REPRESENTATION _SEQUENCE (‘name’, (list of elements: #REPRESENTATION_ITEM), suppressed items (optional): #REPRESENTATION_ITEM, ‘rational’).

• PROCEDURAL_SHAPE_REPRESENTATION. This entity is used to link a creation sequence to a AP203 model. The attributes of this entity are the followings: PROCEDURAL_SHAPE_REPRESENTATION (‘final object’, #PROCEDURAL_SOLID _REPRESENTATION_SEQUENCE, #GEOMETRIC _REPRESENTATION_CONTEXT).

5.3 STEP file creation

The following example illustrates how the STEP AP203 file can be enriched with new Part 108 capabilities. It defines a parallel offset with dimension between two planes. This offset is parameterised.

#1=CARTESIAN_POINT(‘Axis2P3D Location’,(0.,0.,0.)) #2=CARTESIAN_POINT(‘Axis2P3D Location’,(0.,0.,50.)) #3=DIRECTION(‘name’, (0.,0.,1.)); #4=DIRECTION(‘name’, (1.,0.,0.)); #5=DIRECTION(‘name’, (0.,0.,–1.)); #6=DIRECTION(‘name’, (–1.,0.,0.)); #7=AXIS2_PLACEMENT_3D(‘Plane Axis2P3D’,#1,#3,#4); #8=AXIS2_PLACEMENT_3D(‘Plane Axis2P3D’,#2,#5,#6). #9=PLANE(‘Bottom’,#7); #10=PLANE(‘Top’,#8);

#11=PARRALLEL_OFFSET_GEOMETRIC_CONSTRAINT_WITH_DIMENSION(‘PlanePOGC’,‘Height’,(#9), (#10),SURFACE_OFFSET,50.,.F.); #12=FINITE_REAL_INTERVAL(40.,.OPEN.,60., CLOSED); #13=BOUND_MODEL_PARAMETER(‘BParameter’, #12,*,‘Height Parameter’,*); #14=INSTANCE_ATTRIBUTE_REFERENCE(‘GEOMETRIC_MODEL_SCHEMA.PARRALLEL_OFFSET_GEOMETRIC_CONSTRAINT_WITH_DIMENSION.RADIUS’, #11); #14=BOUND_PARAMETER_ENVIRONMENT(#13,#14);

The entities defined in the example does not come from the STEP files processed in test-cases, they are fictive (to simplify the understanding). Each ‘#number’ is a STEP unique identifier used to create relationships between each entity. A plane is defined by a normal axis. The axis of a plane is defined by a point and two coplanar directions. A parallel offset is defined between the two planes and the value of the offset is given ‘50’. This value is associated with a parameter. In this way, the value of the offset is defined as a variable which can be modified by any users of the models. The finite real interval defines the domain of validity of the offset value which is 40 (excluded) to 60 (included).

6 STEP translator specification

6.1 Hypothesis

In this paper, it is assumed that the digitising operation is already done. It is also assumed that geometric data is known for all the geometric features within the point cloud. This data are known using, for example, an automatic methodology for feature recognition (Mohib et al., 2006) or an interactive methodology using a CMM.

In the following section, a translator is presented to convert data of a given geometric feature (a cylinder for example) into the associated STEP feature.

6.2 Use cases diagram The translator has been developed using Unified Modelling Language (UML). First, a use-case diagram (Figure 2) has been developed. This diagram describes the main functions of the translator. Two actors are considered, a CMM technician and a designer.

The CMM technician measured the part to be modelled and informs the translator about the maximal dimension of the original part: this information represents usually the maximum size the CMM can measure and is used to provide boundaries to the part to be modelled.

The Designer uses the translator to transform the original CMM issue into a 3D model based on geometric features. With the human interface included in the translator, the designer enriches the 3D model to define parameters and constraints. Finally, the designer interacts with the STEP translator interface to obtain a STEP file.

8 S. Remy, G. Ducellier, S. Charles and B. Eynard

The translator includes two major functions: • the translation of an original CMM file into a 3D model

(use cases 1–3) • the enrichment of the 3D model generated with

geometric parameters and constraints and the translation into a STEP file (use cases 4 and 5).

As the CMM and STEP entities are different, the transfer from the measured geometry described into the CMM file to a 3D model has been done using a 3D modeller. This 3D modeller enables the user to generate the geometry of the part using the CMM File, to enrich the 3D model with parameters and constraints, and to finally transfer the 3D model into an advanced STEP file enriched with the latest developments in terms of parametric features. The 3D modeller can be seen as a computer aided system for mapping the CMM file and the STEP entities identified in Section 5. Many commercial and open source systems for 3D modelling can be used, depending on their openness and their possibilities in terms of customisation.

6.3 Sequence diagram

Based on the use case diagram above presented, a sequence diagram (Figure 3) has been developed. This diagram

illustrates the dialogs between the actors and the translator. It presents three major sequences:

• The interaction between the CMM technician and the translator. In this sequence the CMM technician provides first the CMM file resulting from the measurement of the real part. Second, the CMM technician provides to the translator parameters needed during the 3D modelling sequence. These parameters concern the boundaries of the part measured and the CMM equipment used for the measurement. They are strongly depending on the used CMM.

• The interaction between the Designer and the translator. In this sequence, the translator assists the designer in the 3D modelling phase. It provides data concerning geometric features measured by the CMM. The 3D modelling process requires human interactions to specify parameters and constraints. These parameters and constraints enrich the geometry and enable the modification of the features created.

Finally, a STEP file integrating Part 108 entities is generated, based on the geometry and on the parameters and constraints specified by the designer.

Figure 2 Use case diagram of the translator

Figure 3 Sequence diagram of the translator

Advanced STEP parameterised and constrained features for Reverse Engineering 9

6.4 Static view of the system From the UML diagrams presenting the functionalities of the translator, a static view (Figure 4) has been developed. This static view presents the entities managed by the application, their links and their multiplicities. As the implementation uses an existing 3D modeller, the static

view has been developed according to the 3D modeller used. Therefore, it may present different structures while using another 3D modeller.

The static view has been simplified due to the size of the global static UML diagram of the application. It presents the 3D modeller and STEP entities relationships.

Figure 4 Static view of the translator

6.6 Implementation of the application

The translator has been implemented using a commercial CAD system. This choice is the consequence of the need for a 3D modeller in order to simplify the 3D visualisation of the object measured with the CMM. Many 3D modellers could have been used but our choice was to develop based on the CATIA V5 APIs for the portability and the simplicity of the code generated. It also integrates required technology to handle parameters, constraints and relations in a object-oriented environment.

The human interface is included in CATIA V5 and supported by Visual Basic for Application (VBA) Technology. The interface reads the CMM file and lists all the geometric features that have been measured. Each of these features can be selected by the designer and automatically modelled in CATIA V5. The designer can enrich the geometry generated by specifying relations or constraints that are not defined in the CMM File. When the retrieving of the geometry is completed, the designer can save the geometric feature using the STEP format.

As an application to illustrate this approach, the following example will enable to understand its different steps. The chosen domain is RE of basic mechanical part. In this case, it is a cross that connects the gear box to the driveshaft of an old convertible 403 by Peugeot.

Reconstruction of old car is a very good benchmark for RE methodology and a current interest for some of the authors of this paper (Remy et al., 2003) (Figure 5). In fact, old cars can provide a lot of aesthetic surfaces and many mechanical parts … and no CAD models are available. As Remy et al. (2003) presents an approach to deal interactively with aesthetic surfaces digitised with a laser sensor, the approach introduced in the paper is more dedicated to the RE of mechanical part. As said before, the part that has been chosen for this example is a cross that enables to transmit a torque from a gear box to a driveshaft using a Cardan joint. Regarding of the methodology described in this paper, only the geometric primitives (plane, cylinder, sphere, cone …) are considered.

10 S. Remy, G. Ducellier, S. Charles and B. Eynard

Figure 5 Reverse Engineering of a radiator cap

In every RE application, the first step is the acquisition of the data that are needed to rebuild the part. The different tools used are the following: A CMM by Wenzel with a PH10 head by Renishaw and a touch probe. On this part,

10 planes and 6 cylinders are digitised. One by one, these features are measured.

For each plane, the software carrying out the measurement asks for a minimum of 4 points. It provides the normal vector of the plane and a point that belongs to the plane (the centre of mass of the measured points). For each cylinder, it is a minimum of 8 points that are needed by the software. This one provides the direction of the axis, a point that belongs to this axis (the projection on this axis of the centre of mass of the measured points) and the radius. Other data like planarity or cylindricity are given but they are not used.

After the measurement, the different data provided by the CMM are used to build directly the STEP file of each feature. Here, CATIA V5 is only used as a viewer. The screenshots on Figure 6 propose a plane and a cylinder. The STEP files including parameterised features are directly built using the data provided by the CMM.

Figure 6 The whole approach

In this example, the 16 features are measured and 16 STEP files are built. The 16 recovered surfaces were merged and cut through Boolean operations to create the cardan joint shell. Then, a solid model was created by filling the shell. After that, constraints and parameters were added to model functional specificities of the part. Finally, the complete CAD model is built as a unique STEP file based on parameterised features (Figure 6).

7 Conclusion

The paper has described some of the capabilities of ISO 10303-108, a new part of the ISO 10303 international

standard (known as STEP). It allows the inclusion of parameterised and constrained data in the inter-system exchange of explicit models in order to share models based on parametric 3D features. The preliminary application areas for such parameters and constraints in CAD models are 2D sketches or profiles, inter-feature relationships and inter-part relationships in 3D assembly models. The Part 108 capabilities were assessed through its implementation in a RE application.

Models created by standard RE approaches are often frozen and does not contain constraints and parameters (which are essential to represent design intent). The results of the developed case-tests demonstrated that

Advanced STEP parameterised and constrained features for Reverse Engineering 11

implementation of Part 108 in this field could fill these gaps. These case-tests have successfully show efficiency of the Part 108 for enabling RE of parts based on STEP parameterised and constrained features.

The future challenges of this work will be to apply the proposed approach on RE of complex parts composed of numerous features. In this future work, it will be necessary to categorise the features into geometric family calling necessary objects of the translator. The final stage of this work will be dedicated to the assessment of the STEP based RE approach in the context of parts assemblies in order to complete the approach with more complex CAD modelling.

References AUTHOR PLEASE SUPPLY PAGE RANGE FOR ALL

JOURNALS. Chivate, P.N. and Jablokow, A.G. (1993) ‘Solid-model generation

from measured point data’, Computer-Aided Design, Vol. 25. Choi, B.K., Shin, H.Y., Yoon, Y.I. and Lee, J.W. (1988)

‘Triangulation of scattered data in 3D space’, Computer-Aided Design, Vol. 20.

Cignoni, P., Montani, C. and Scopigno, R. (1998) ‘DeWall. A fast divide and conquer Delaunay triangulation algorithm’, Computer-Aided Design, Vol. 30.

Fang, T.P. and Piegl, L. (1995) ‘Delaunay triangulation in three dimensions’, IEEE Computer Graphics and Applications, Vol. 15.

Fowler, J. (1995) STEP for Data Management Exchange and Sharing, Technology Appraisals.

Galetto, M. and Vezzetti, E. (2006) ‘Reverse engineering of free-form surfaces: a methodology for threshold definition in selective sampling’, International Journal of Machine Tools and Manufacture, Vol. 46.

Hardwick, M., Morris, K.C., Spooner, D.L., Rando, T. and Denno, P. (2000) ‘Lessons learned developing protocols for the industrial virtual enterprise’, Computer-Aided Design, Vol. 32.

Hoffman, R. and Jain, K. (1987) ‘Segmentation and classification of range images’, IEEE Pattern Analysis and Machine Intelligence, Vol. 9.

Hoppe, H., DeRose, T., Duchamp, T., McDonald, J. and Stuetzle, W. (1992) ‘Surface reconstruction from unorganized points’, Computer Graphics – Proceeding of SIGGRAPH. AUTHOR PLEASE CHECK IF THE AUTHOR NAME IS ‘Hope’ OR ‘Hoppe’.

ISO 10303-1 (1994) ‘Industrial automation systems and integration – product data representation and exchange – Part 1: overview and fundamental principles’, International Standard Organization, Geneva. AUTHOR PLEASE CITE THIS REFERENCE IN TEXT.

ISO 10303-108 (2004) ‘Industrial automation systems and integration – product data representation and exchange – Part 108: integrated application resource: Parameterization and constraints for explicit geometric product models’, International Standard Organization, Geneva.

Lai, J. and Lai, H. (2006) ‘Repairing triangular meshes for reverse engineering applications’, Advances in Engineering Software, Vol. 37.

Lawson, C.L. (1977) Software for C1 Surface Interpolation, Mathematical Software III, Academic Press. AUTHOR PLEASE SUPPLY LOCATION.

Mohib, A., Remy, S. and El Maraghy, H. (2006) ‘Recognition of geometric primitives using medial axis transform’, Proceedings of the CIRP Design Seminar, Kananaskis, Canada.

Piegl, L. and Tiller, W. (1995) The NURBS Book, Springer, Berlin. Pratt, M.J. (2004) ‘A new ISO 10303 (STEP) resource for

modeling parameterization and constraints’, Journal of Computing and Information Science in Engineering, Vol. 4.

Pratt, M.J., Anderson, B.D. and Ranger, T. (2005) ‘Towards the standardized exchange of parameterized feature-based CAD models’, Computer-Aided Design, Vol. 37.

Remy, S., Ris, G., Nartz, O. and Bernard, A. (2003) ‘Reverse engineering of a 1935 Delahaye radiator cap’, Proceedings of European Virtual Engineering Network Conference 2003, Dublin, Ireland.

Sakar, B. and Menq, Ch. (1991) ‘Smooth surface approximation and reverse engineering’, Computer-Aided Design, Vol. 23.

Scheder, H. (1995) ‘Product data integration – Needs and requirements from industry’, Proceedings of European Product Data Technology Days’95 – Using STEP in Industry, Munchen, Germany. AUTHOR PLEASE CHECK IF THE AUTHOR NAME is ‘Sheder’ OR ‘Scheder’.

Soucy, M. and Laurendeau, D. (1995) ‘A general surface approach to the integration of a set of range views’, IEEE Pattern Analysis and Machine Intelligence, Vol. 17.

Spooner, S. and Hardwick, M. (1997) ‘Using views for product data exchange’, IEEE – Computer Graphics and Applications, Vol. 17.

Urk, G. and Levoy, M. (1994) ‘Zippered polygon meshes from range images’, Proceedings of SIGGRAPH’94. AUTHOR PLEASE SUPPLY LOCATION AND PAGE RANGE.

Varady, T., Martin, R. and Cox, J. (1997) ‘Reverse engineering of geometric models-an introduction’, Computer-Aided Design, Vol. 29.

Weir, D.J., Milroy, M.J., Bradley, C. and Vickers, G.W. (1996) ‘Reverse engineering physical models employing wrap-around B-spline surfaces and quadrics’, Proceedings of the Institution of Mechanical Engineers – Part B, Vol. 210.

Yeh, S.C. and You, C.F. (2000) ‘Implementation of STEP-based product data exchange and sharing’, Concurrent Engineering Research an Applications, Vol. 8.

Zhang, Y., Zhang, C. and Wang, B. (1998) ‘Interoperation of STEP application protocols for product data management’, Concurrent Engineering Research and Applications, Vol. 6.