On this day in 1974, the first Universal Product Code was scanned at a supermarket cash register. The UPC bar code system was originally invented specifically for grocery stores, to speed check-out and help them keep better track of their inventory, but it proved so successful that it spread quickly to other retailers. The first patent for a bar code went to N. Joseph Woodland and Bernard Silver in 1952. Silver, a graduate student at Drexel Institute of Technology, overheard the president of the local food chain, Food Fair, asking one of the deans to research a system to automatically read product information during checkout. He told his friend Woodland and they began work on the idea. The first system used ultraviolet ink, which was expensive and faded quickly.
Convinced that the system was workable with further development, Woodland left Drexel, moved into his father's apartment in Florida, and continued working on the system. His next inspiration came from Morse code, and he formed his first barcode from sand on the beach. "I just extended the dots and dashes downwards and made narrow lines and wide lines out of them." To read them, he adapted technology from optical soundtracks in movies, using a 500-watt incandescent light bulb shining through the paper onto a photomultiplier tube from a movie projector on the far side. He later decided that the system would work better if it were printed as a circle instead of a line, allowing it to be scanned in any direction.
In 1951, Woodland moved to IBM and continually tried to interest them in developing the system. The company eventually commissioned a report on the idea, which concluded that it was both feasible and interesting, but that processing the resulting information would require equipment that was some time off in the future. IBM offered to buy the patent, but their offer was rejected in favor of one from Philco, which then sold it to RCA.
In 1966, the National Association of Food Chains (NAFC) held a meeting on the idea of automated checkout systems. RCA, who had purchased the rights to the original Woodland patent, attended the meeting and initiated an internal project to develop a system based on the bullseye code. The Kroger grocery chain volunteered to test it.
In the mid-1970s, the NAFC established the Ad-Hoc Committee for U.S. Supermarkets on a Uniform Grocery-Product Code to set guidelines for barcode development. In addition, it created a symbol-selection subcommittee to help standardize the approach. They developed a standardized 11-digit code for identifying products. The committee then sent out a contract tender to develop a barcode system to print and read the code. The request went to Singer, National Cash Register (NCR), Litton Industries, RCA, Pitney-Bowes, IBM and many others. A wide variety of barcode approaches was studied, including linear codes, RCA's bullseye concentric circle code, starburst patterns and others.
In the spring of 1971, RCA demonstrated their bullseye code at another industry meeting. IBM executives at the meeting noticed the crowds at the RCA booth and immediately developed their own system. Woodland, still employed by IBM, established a new facility in Raleigh-Durham Research Triangle Park to lead development.
In July 1972, RCA began an 18-month test in a Kroger store in Cincinnati. Barcodes were printed on small pieces of adhesive paper, and attached by hand by store employees when they were adding price tags. The code proved to have a serious problem; the printers would sometimes smear ink, rendering the code unreadable in most orientations. However, a linear code, like the one being developed by Woodland at IBM, was printed in the direction of the stripes, so extra ink would simply make the code taller while remaining readable. So on 3 April 1973, the IBM UPC was selected as the NAFC standard.
The IBM 3660 included a digital cash register and checkout scanner, and the grocery industry, which had been collaborating with IBM on the invention, began requiring its suppliers to start putting bar codes on their packaging. There were no cost savings for a grocery to use it, unless at least 70% of the grocery's products had the barcode printed on the product by the manufacturer. IBM projected that 75% would be needed in 1975. Yet, although this was achieved, there were still scanning machines in fewer than 200 grocery stores by 1977. On the other hand, experience with barcode scanning in those stores revealed additional benefits. The detailed sales information acquired by the new systems allowed greater responsiveness to customer habits, needs and preferences, and to collect information on personal buying habits to target advertising.
The first scan was made at a Marsh's Supermarket in Troy, Ohio, which had agreed to serve as a test facility for the new technology, and the first item scanned was a pack of Wrigley's Juicy Fruit Gum. There's no significance to gum being the first item scanned; it just happened to be the first thing pulled from the cart. That pack of gum is on display at the Smithsonian's National Museum of American History in Washington, D.C.
The technology itself is remarkably clever and simple. Think of what it has to do -- scan a code of any size on any color of packaging held in any orientation without taking forever to do it. There have been a variety of technologies developed to do this but the most common is based on a helium-neon laser.
The first problem is how to see the code in the first place. Through either a moving mirror or a rotating prism, the laser scans across the package. Another mirror directs the reflected light to a photodiode, an electrical device that switches on or off depending on the intensity of the light shining on it.
The second problem is how to prevent the system from spurious signals from extraneous light shining on it. This is achieved by causing the brightness of the laser to vary in a specific pattern. Any light that does not match the modulation pattern is ignored.
The third problem is how to pick the barcode out from all the other crap on the package. This is accomplished by having special beginning and ending bars. These are the longer ones on the sides of the numerical code. If the machine sees both of those, anything in between is a barcode and if it doesn't those signals are ignored.
For a lot of applications, this is all that is needed. The handheld readers at Home Depot simply scan horizontally and you have to hold the barcode in that orientation.
This would be too much of a delay at the grocery store, and the orientation problem is solved in a couple of ways by having the laser scan in two dimensions. Some of them emit a starburst pattern while others create a Lissajous pattern (see below). In both cases, the idea is that one of the scan lines (for the case of the starburst) will cross all the bars in the code.
A Lissajous pattern is generated by controlling the scan mirror in two dimensions with different frequencies in each dimension. Since you can vary the frequencies, you can change the number of loops in the pattern and cause it to rotate. Eventually a scan line will cross the bar code.
Here is a nice mechanical model of how they work that makes the process visual:
Here is the same thing, electronically generated, that show how the pattern can be rotated to scan:
And here are the sorts of complex patterns that can be generated, with audio so you can hear the two signals being used:
With one of those, you would not have to wait very long for the barcode to be detected.
Comments
Post a Comment