Curb Redundancy, Maximize Reuse12 Sep, 2007 By: Jeffrey Rowe
Geometric shape search engines continue to evolve as the potential uses grow.
As popular as text search engines, such as Google, have become over the past few years, I wondered if something analogous was occurring in the MCAD world with the zillions of geometry-based parts that are continually generated. As it turns out, several are currently available on the market. Although they have not received nearly the notoriety of their text search engine counterparts, they will.
The shape search engines currently on the market differ somewhat in their underlying technology and are used for different purposes. Their potential uses continue to be discovered. For example, you could use a shape search engine to ensure that no duplicates of a design are hidden away in massive enterprise MCAD part databases. Designers could also use a shape search engine for identifying and reviewing design alternatives. Since, theoretically, shape search engines are MCAD package independent, I fully expect them to migrate to and become integrated with virtually all CAD tools in the near future. Doing so obviously will expand their potential user base within enterprises.
Let’s look at a couple of the shape search engines currently available and the companies that developed them.
iSEEK’s CADSEEK began with CAD and shape-based search algorithms that were developed at Iowa State University. Initially the company’s founders recognized that virtually all manufacturers struggled to manage their engineering information. CAD files, the core component of engineering data, were typically dispersed across an organization. Different versions of parts were often mixed together, and finding the desired file could be a time-consuming -- and time-wasting -- process. This situation often forced engineers to (re)design a part from scratch rather than spend time trying to find the existing part they need.
iSEEK's development of CADSEEK was intended to eliminate this problem as well as the issue of part redundancy. Simply put, the CADSEEK Search Engine is a geometric-shape-based search module for finding and comparing parts or assemblies for similarity.
CADSEEK's surface-based algorithms can perform similarity searches, including finding mating parts, finding parts used in higher level assemblies and within a given percentage, and finding the casting or forging from which a given part is manufactured.
The Visual Navigator tool uses the CADSEEK search engine algorithm to provide a complete visualization of CAD database files. These files are clustered by similarity and displayed as parts "billboards," which vary in size depending on the number off files that make up the cluster. Users can group them by material, corporate division, project, supplier, and other details. Parts can also be selected from the Visual Navigator to initiate a similar part search.
A module presents 3D information of the Visual Navigator in 2D thumbnails with what the company calls the Digital Catalog. It presents the thumbnail images in a "page" format so users can browse to find parts of a particular shape. The Digital Catalog is a thin client that operates in a Web browser.
The Digital Catalog also utilizes metadata, text extracted from individual CAD files such as material type, dimensional data, thread pitch, designer name, supplier name and location, and cost. With CADSEEK’s Digital Catalog, searches combine geometric shape and metadata, a capability that is particularly useful because of the versatility of the search input parameters.
I reviewed this product briefly in conjunction with SolidWorks 2007 when it was then known as 3DSearchIT for SolidWorks. 3DSearchIT is a shape (geometry)-based search engine that is CAD independent and, according to the company, can work with any 3D file format.
When asked about the SolidWorks product, Rajnish Bharti, senior product manager for Geometric Software Solutions , said, “We are seeking customization and implementation opportunities in the 3DSearch domain. Going forward, our focus will be on customizing the technology for individual companies' needs. The technology is CAD format agnostic and can be easily tweaked to support formats other then SolidWorks. 3DSearchIt for SolidWorks is an out-of-the-box product based on our 3DSearch technology and was created with the intent to provide a technical preview of the possibilities. Moving forward, however, we will not continue to invest in this product.”
So Geometric's 3DSearchIT technology will remain a modeler-independent, geometric 3D-shape-based encoding and search library running under a client-server architecture. By representing raw 3D data, 3DSearchIT’s technology provides an efficient tool for 3D information retrieval and reuse at just about any scale. 3DSearchIT can be integrated with any PDM system, and its Web-based part catalogs provide the ability to reuse existing product data.
|Schematic representation of 3DSearchIT server-client architecture.|
3DSearchIT technology can help users find parts based on 3D geometry using recognition tools that extract features and their parameters from solid models. It has applications in modeling, design, finite-element analysis, machining, process planning, and cost estimation. Feature recognition is the ability to automatically or interactively identify and group topological entities, such as faces in a boundary representation (B-rep) solid model, into functionally significant features such as holes, slots, pockets, fillets, ribs, and more. The features that are not recognized automatically can be identified using the interactive feature recognition capabilities. Feature recognition is what makes smart solids out of dumb solids, providing information that can be used for other purposes.
Next time I’ll discuss Geolus Search, a technology that UGS PLM Software recently acquired. I’ll also cover some technical details on how the various shape search engines actually work because most share some common characteristics. As I said earlier, shape search engines are just now coming into their own, and their potential is huge.