18 - Feature Recognition for Smart Design and Manufacture
Submitting Institutions
Heriot-Watt University,
University of EdinburghUnit of Assessment
General EngineeringSummary Impact Type
TechnologicalResearch Subject Area(s)
Information and Computing Sciences: Artificial Intelligence and Image Processing, Computation Theory and Mathematics, Information Systems
Summary of the impact
In 1997 ERPE invented a novel automatic machining feature recognition
technology which has been incorporated into the Pathtrace EdgeCAM Solid
Machinist Computer Aided Manufacture (CAM) package, now owned by Planit
plc. EdgeCAM is considered as one of the leading independent solid
machinist CAM package, with 10 - 15% of the world market. Related ERPE
feature recognition in shape representation and characterisation has
enabled the design of a 3D shape browser for product data management
systems. Commercialised in 2005 as ShapeSpace with £0.7M current market
value, for application to the parts industry in automotive markets, it has
attracted the US Actify Inc., as an equity sharing partner to aid
ShapeSpace to access worldwide markets.
Underpinning research
The ERPE research team comprises Profs: Corney (left 2007); Clark
(retired 2003); Lim, Senior Lecturer Mill, with Prof Taylor from Computer
Science (all in post throughout the period unless stated). Former
lecturers are Salmon and Sherlock, and former PDRAs: Little; Tuttle; and
Sormaz.
Advances in feature recognition for parts machining was motivated by
industry's need to create a faster, more efficient and robust Computer
Aided Manufacture (CAM) system. Corney, Clark, Little, Tuttle and Sormaz
first applied computational geometry with graph theory to identify
numerically controlled (NC) machining features from solid models [1]. This
early implementation was novel but was limited to only prismatic shapes.
Lim then refined and extended this to curved and complex surface
geometries [3].
The ability to automatically find machining features on curved and
prismatic components interested Pathtrace (developers of EdgeCAM) and
subsequently resulted in a knowledge transfer programme. Little (1998)
moved to Pathtrace to implement the first development as an AutoCAD
plugin. Lim (1999) then worked in Pathtrace to demonstrate the
effectiveness and robustness of the new automatic feature recognition
algorithms and incorporated them into EdgeCAM.
The research was incorporated into Pathtrace's roadmaps towards achieving
highly intelligent CAM systems. The novel design and packaging of the
algorithms showed that it is highly scalable, tractable and adaptable to
function as the core architecture in solids machinists' software systems.
This is extremely important as feature finding is today a key element when
preparing solid parts for manufacturing [2].
A related commercialised research activity, ShapeSpace Ltd, founded by
Sherlock and Mill was originally aimed at developing new geometric
reasoning algorithms to support design activities and aid the description
of parts to assist identifying their location within a large stock or
database. This researched the development of a 3D browser that enabled
efficient searching for models and other design data with the patent
application now filed in 2011 as US 13/698809, with a priority date of
18/05/2010.
Initial research [4] (Salmon to 2000) was based on the development of
high level design tools which incorporated a rich set of information in
addition to geometry. These 3D feature enhanced CAD modellers were aimed
at improving general company activities such as management and
manufacturing functions. This led to research that was concerned with the
representation of parts and their associated data [5], and developing more
flexible approaches to dealing with the associated manufacturing
information. This in turn led on to exploring the specific problems of
finding product-related data in large company file systems and intranets.
Collaborating with partners in Europe and Corney's group permitted
superior algorithms to be developed that would allow engineers to more
efficiently search for complex design data in a flexible way. A major
innovation was the development of shape based search methods to support
human computer interaction [6] with Taylor.
The foundations of this research can be traced back to Mill's awards
(GR/K53659/01, GR/J40713/01, GR/F92312/01 totalling £272k in association
with GEC Marconi, Rolls Royce, BAE Systems and Logica), the EC/BRITE-EURAM
award SESAME grant, £235k, with Spatial Technology Ltd./Strassle
GmbH/MandelliSpA which supported the basic engineering work in shape
representation and characterisation, particularly in feature-based design
and shape optimisation.
There is now a centre of expertise based around the Edinburgh offices of
ShapeSpace, with its US partner Actify Inc., and through ongoing
collaborations with the ERPE and Strathclyde.
References to the research
References identified with * are those which best indicate the quality of
the underpinning research.
[1] *Tuttle, R., Little, G., Corney, J, and Clark, D.E.R., "Feature
recognition for NC part programming", Computers in Industry, Vol.
35, No. 3, pp. 275-286, April 1998 DOI: 10.1016/S0166-3615(97)00089-4.
22 Google Scholar (GS) citations.
This paper reports the automated recognition of manufacturing features on
solid models of polyhedral objects to support the generation of NC machine
code for process planning and manufacturing.
[2] *Lim, T., Corney, J. and Clark, D.E.R, "Exact Tool Sizing for
Feature Accessibility", International Journal of Advanced
Manufacturing Technology, Vol. 16, No. 11, pp. 791-802, 2000. DOI:10.1007/s001700070013.
26 GS citations.
This paper presents an algorithm for calculating the volume of a
2D-profile, accessible by a given diameter of milling cutter. Exact
results are obtained despite simplicity of this procedure.
[3] Lim, T., Corney, J. and Clark, D.E.R, "Laminae-Based Feature
Recognition", IEEE Transactions on Pattern Analysis and Machine
Intelligence, Vol. 23, No. 9, pp. 1043-1048, 2001. DOI:10.1109/34.955117
This paper presents a novel approach to recognizing shape features on
geometric models using a network of adjacent 2D-laminae or bounded
surfaces.
[4] Naish, J., Mill, F. and Salmon, J, "An Industry-based Study of
Cutting Process Capability: Representation Requirements for an
Integrated Simultaneous Workstation", Institute of Industrial
Engineers (IIE) Transactions, Vol. 29, No. 7, pp. 573-584, 1997. DOI:10.1023/A:1018501530306
This paper uses a library of volumetric features to build 3D models that
are typical of those created by cutting and milling processes.
[5] Baron, P., Fisher, R., Mill, F., Sherlock, A., and Tuson, A., "A
voxel based representation for evolutionary shape optimization",
Artificial Intelligence for Engineering Design, Analysis and
Manufacturing, Vol. 13, No. 3. pp. 145-156, 1999. DOI:10.1017/S0890060499133031.
12 GS citations.
This shows how any arbitrary part can be approximated with the voxel based
representation. This can be used as a suitable representation for a
genetic algorithm based shape optimization.
[6] *Clark, D.E., Corney, J.R., Mill, F., Rea, H., Sherlock, A. and
Taylor N.K. "Benchmarking Shape Signatures against Human Perceptions of
Geometric Similarity", Computer Aided Design Journal, Vol. 38, No.
9, pp. 1038-1051, 2006. DOI: 10.1016/j.cad.2006.05.003.
10 GS citations. Methods that have been developed to automatically
characterize 3D shapes in large databases are compared to methods based on
human judgment to test the researchers' algorithms.
Further Grant funding:
[G1] SERC/ACME, £130k, "Genetic Algorithms in Manufacturing
Engineering", 1993 - 1996, with Rolls Royce/British
Aerospace/University of Sussex.
[G2] ESPRC, GR/K48020/01, £268k (Pre-FEC), Corney (PI) "Feature
Recognition For CNC Part-Programming", 1995 - 1998
[G3] ESPRC, GR/N21307/01, £176k (Pre-FEC), Corney (PI), "Part Sourcing
In A Global Market", 2000 - 2003
[G4] ESPRC, GR/R35285/01, £179k (Pre-FEC), Corney (PI) "Assembly Based
Rapid Prototyping" 2001 - 2003
Details of the impact
ERPE researchers have advanced and contributed to the science of
automatic feature recognition.
This research was critical to the development of the feature finder
functionality now incorporated in the EdgeCAM "Solid Machinist" module and
brought the product to market significantly quicker than would otherwise
have done [S5]. According to the Software Development Manager, Vero UK
(Planit's parent company), "This work brought Solid Machinist to market
quicker and some of the algorithms are still in use. The technology was
part of the reason why Planit PLC bought Pathtrace in 2006, as the vital
part of the functionality in EdgeCAM was the principal product of
Pathtrace" [S1]. Solid Machinist is used by customers all around the
world for milling and turning of components, modelled in various different
CAD formats.
EdgeCAM Solid Machinist uses automatic feature recognition [1, 3]
to interrogate the solid model and quickly identify machinable features.
EdgeCAM then offers the user the most appropriate tooling and machining
strategy to generate accurate toolpaths [2]. "The research developed
collaboratively with ERPE provided the building blocks of the software
used by our core customers in industries such as aerospace, automotive,
energy and medical sectors. From a position where 7 years ago, 20% of
our customers used the software, to approximately 65% of our customers
using the software now demonstrates that the integrity of the
collaborative research means that the technology meets customer needs
more closely. Without strong University collaboration, developing
sustainable R & D is very difficult to justify financially for a
company like EdgeCAM. While it is considered one of the world's leading
independent solid machinist CAM package, with 10 - 15% of the world
market, the demands on the business mean that there is little capacity
to undertake such high risk work." Senior Manager at EdgeCAM. [S2]
The related company ShapeSpace Ltd., was developed to provide
leading edge tools to design engineers and to those concerned with the
management of design data. With Scottish Enterprise and Royal Society of
Edinburgh support in 2005, through an Enterprise Fellow award, Sherlock
launched ShapeSpace. Private funding and a SMART award led to the
development of search software for engineering and manufacturing companies
in 2008.
A specific problem addressed is the location of product-related data in
large file systems and intranets. The ERPE 3D shape browser was thus
designed [4, 5] to allow engineers to search for product data in their
management systems or databases via the associated part shapes. This
enables design re-use, better file management through the identification
of duplicate and similar parts. Additionally recent collaborations have
led to the development of tools that assist managers to better view the
state of the total design effort for new products.
"One of the most costly overheads of 3D CAD design and modeling is the
creation of stock parts. ShapeSpace makes searching for these, sometimes
mislaid, parts a fast and intuitive process and has already saved me
hours spent recreating already existing models...We model shapes in 3D
so what better way to search for them than in 3D!" Senior Design
Draughtsman, Chemring EOD Ltd., [S3].
These tools have led to directly sellable products that are now offered
by the company and its long term development plan is based on ERPE
collaboration. In 2008 the company's strategy changed to develop expertise
which would make them recognized as world experts in part searching
technologies and in the representation of complex product data. This led
to the ShapeSpace partnering with Actify Inc., (a San Francisco based
software company) on an initial two year contract with Jaguar Land Rover
for the supply of product status tracking software for all new vehicle
programs in development.
In October 2012, the formal investment and equity sharing arrangement
with Actify Inc., http://www.actify.co.uk/resources/press-releases/812-2/,
led to the opening of a joint office [S6] with a new period of growth to
access to new markets, especially automotive industries in the US and
Germany, and the continued development of new products around data
analytics and business intelligence for discrete manufacturing. Currently,
6 people are employed in the Edinburgh office. ShapeSpace current market
value, calculated on recent share trades is around £0.7M, following the
significant investment by Actify.
The new products, which enhance the original technology, in the area of
data analytics for product data which result from the 2012 SMART award,
have seen significant traction in the market both in the UK and globally.
Current customers include Jaguar Land Rover, Weir Group, CSC, JCB, Lucy
Switchgear in the UK and Exterior Systems Engineering, Benteler Automotive
Corporation, Magna Inc, Complete Prototype Services, Inc., Inalfa Inc,
Dienamic Tooling Systems, Inc and Autotek Mexico, S.A. in North America.
"The research at the University of Edinburgh on geometric reasoning
for shape recognition was instrumental in the development of the
technology that underpins ShapeSpace's growth in the area of data
analytics for engineering and manufacturing industries. The University's
support of Andrew Sherlock's academic career, which led to him securing
the Royal Society of Edinburgh-Scottish Enterprise funded Enterprise
Fellowship, very successfully enabled the formation of our company,
ShapeSpace." Senior Software Engineer, ShapeSpace [S4].
Sources to corroborate the impact
[S1] Software Development Manager, Vero UK, see comments included in
Section 4.
[S2] Senior Manager, EdgeCAM, see comments included in Section 4.
[S3] Senior Design Draughtsman, Chemring EOD, Ltd., see comments included
in Section 4
[S4] Senior Software Engineer, ShapeSpace, Ltd., see comments included in
Section 4.
[S5] Sales Manager, Pathtrace will confirm that this work was critical to
the development of the feature finder functionality incorporated in
EdgeCAM "Solid Machinist" module and brought the product to market quicker
than would otherwise have done. Solid Machinist is used by customers all
around the world for milling and turning of components, modelled in CAD
formats.
[S6] http://www.actify.co.uk/resources/press-releases/actify-opens-new-office-in-edinburgh-scotland/
details the Actify relationship with Shapespace.