Open-ended brainstorming and debate. They've considered whether they should go for a Macy's-style department store, a Wal-Mart-style telemarketing list supercenter, or even an electronics store. They had an idea, which was later abandoned, of making the store two floors, letting Amazon's disk-shaped warehouse robots assemble orders on the top floor, and then using conveyor belts and robots to deliver the goods to vehicles waiting for customers downstairs. After a few months, Kumar, Puerini, and their colleagues had to admit that most stores in the real world were already working pretty well, with one notable exception: supermarkets.
Supermarket checkouts are a nuisance, the average American buys groceries twice a week, and their experience waiting at the checkout telemarketing list can be summed up as—in the words of Amazon’s Type A disruptor cracking team—inefficient Tired offline shopping. "We realize that there are many benefits to shopping in a brick-and-mortar store, but waiting in line is not one of them," Kumar said. While many companies are trying to solve this hassle, Apple has employees walking around Apple Stores with card swipe machines in hand, while Bingo Box in China uses RFID chips attached to product packaging for self-checkout, and the IHM team hopes to completely eliminate bottlenecks. There's a tradition at Amazon that to make sure the team works backwards to what the customer wants, they start with a press release, or in Amazon's parlance, a "PR Faq," announcing that they'll be opening a store without a checkout. Then start working on the actual technology to turn what is released into reality.
It turned out to be easier said than done, and much more expensive than expected. To figure out who would buy what at a store without telemarketing list a checkout, IHM engineers considered using RFID to track customers' phones as they walked through the aisles and scan their faces using facial recognition technology. They also discussed requiring customers to quickly scan a QR code when selecting something, but while that would make Amazon's job easier, those actions might still seem odd or unnatural to customers. In the end, they decided to use computer vision, a relatively new technology that uses digital cameras and computers alone to identify items by their visual appearance without the need for any special tracking chips or codes.