Vision and auto ID
Article Type: Viewpoint From: Assembly Automation, Volume 29, Issue 2
The importance of identification as a function of vision was highlighted in 2008 by the sale of the Siemens vision activities to Microscan, a US company which had made its name with miniature bar-code readers. Those of us who follow the evolution of vision companies will recall, from the distant past, that the Siemens US activity had been RVSI which in turn had absorbed Acuity, formed initially by the merger of Automatix and Itran. The latter had been pioneers of 2D matrix coding!
Again taking a historical view, the Stuttgart Vision exhibition, which now takes place every November at the new Fairgrounds at Stuttgart Airport, was founded over 20 years ago in nearby Sindelfingen as “Ident Vision,”indeed “Ident” alone had been started the previous year. The “Ident”part of the title was dropped several years ago but the official title remains the “International Trade Fair for Machine Vision and Identification Technologies,” though one no longer sees RFID and bar-code printing products there as in the early days.
Looking at the generally acknowledged world leading machine vision company,Cognex, it was the problem of character recognition for identification purposes that drove the development of their pattern-matching algorithm, a fast implementation of normalised correlation, which turned them into a profitable company in the latter part of the 1980s, though the real cause of success was the superior performance the algorithm gave in finding fiduciary marks in the semiconductor and PCB manufacturing industries.
It has become traditional to analyse the market for vision systems into three very broad application sectors – inspection, recognition (identification either by codes, characters, or even the shape of the part itself), and guidance. In the field of automatic assembly, all three applications have a role. If you want your assembly operation to continue smoothly for hours without human intervention, it is wise to 100 per cent inspect all the components used so that a “rogue” item such as a wrongly threaded bolt or an oversize component will not jam the system until it is manually removed. During the operation, guidance may well be necessary – a classic early application of“robot guidance” was in the robotic mounting of wheels onto cars on an assembly line – any motorist who has changed a wheel knows that it is essential to line up the holes in the wheel with their corresponding holes or studs on the axle stubs before attempting to mount the wheel. After an operation has been performed automatically, it will be wise to inspect automatically to check that the operation has been successful and that the product is either fit for despatch or for a successive operation without risk of adding value to an already defective product.
What about identification? A long-established example lies in those sectors of the automotive industry where cars are literally “built to order”and the customer can specify the exact trim choices, tyres, and so on. It is clearly essential to identify the body uniquely and automatically as it progresses down the assembly line, and while RFID is used on some lines vision has been adopted on others, for instance using a 2D matrix of holes in a place which will eventually be concealed, so that the code can be read even after the entire body has been painted.
These days identification is perhaps more important for traceability of components. If it becomes necessary to recall a batch of cars because a safety-critical component has been found faulty in several instances, the knowledge of exactly which cars had that batch built into them can save an enormous amount of money. One German supplier of automated assembly systems to the automotive industry was explaining at Automatica in 2008 that their latest project required them to not only test the assembled item after assembly,but they also had to include facilities for storing that test data along with the identity of each part used, for as long as 10 years after manufacture!
Don Bragginsbased at Machine Vision Systems Consultancy, Royston, UK
