August 5, 2004

Intelligent Cart Brings 'Jetsons'-Style Shopping

By Evan Schuman, eWEEK

Envision opening a refrigerator, making a tuna-fish sandwich and using the last of the mayonnaise. You scan the empty jar on the freezer door for two seconds and then recycle the sticky jar.

Three days later, you walk into your local grocery store and grab a cart, which displays a list of everything you need. Some of the items came from a list you e-mailed to your store, others are things the cart thinks you're probably out of (you haven't purchased milk in two weeks and your last portion had an expiration date of eight days ago), and some are things that your refrigerator told your shopping cart about.

Your cart uses a navigation system to literally direct you to your items, flashing ads along the way based on your particular buying pattern. If you buy two tubes of toothpaste, it can decide to give you a special discount on the third right there, an offer it will not make to the next customer because you have its intended three-child household.

You look on your list and see that a family member wants a particular flavor of a specific brand of canned soup. The cart takes you to the relevant part of that aisle, but all you see is a sea of cans stacked 30 deep. The flavor you seek is either buried deep behind the front can or it isn't there.

You don't have to start shifting cans. The tablet in the cart checks inventory and reports that the store has sold out of that flavor. Or perhaps it reports that there is only one can left, and by moving the tablet by the cans, it points to the correct row and indicates that it is 11 cans back.

As you put in another can of soup for yourself, the tablet beeps and alerts you that it is high-sodium and it knows you're on a low-sodium diet. The tablet also knows of your son's peanut allergy and will similarly flag any items you purchase that have peanuts anywhere on its ingredient list.

You're about to head out when the tablet beeps again, this time to tell you that your one pound of thinly sliced roast beef is ready at the deli counter. When you arrived at the store, the tablet beamed the order to them directly while you shopped, and then they beamed a message back when it was finished.

There's no need for a checkout lane for you, as the cart charges your credit card for each item the instant you place it in the cart. Don't worry if the produce isn't barcoded, as an IBM device scan reviews color, texture and possibly smell to distinguish between a Granny Smith and a Golden Delicious apple or between a plantain and a banana.

Most of the technology just described is not hypothetical or prototype-only. It's deployed somewhere in the retail world today, mostly in limited test mode. The other items are ready to be deployed in the next rev of various vendors' self-checkout products.

Some concepts that are on the purely theoretical level include inch-by-inch cart tracking, so that retailers could learn which shoppers paused at which displays and then took which action. How many people looked at certain products, read their labels and then put them back?

The system's security method is less restrictive than the nonmobile self-checkout method, which is why it's supplemented by periodic, full-cart re-scans.

Dusty Lutz, product manager for NCR FastLane by NCR Corp., said he has concerns about how consumers will feel about those security audits. "That's a pretty punitive security system," he said, pointing to the randomness and that the entire cart needs to be emptied and re-scanned.

Mike Grimes is vice president of sales and marketing at Cuesol Inc., a leading smart-cart vendor. Grimes said he disagrees that the random audits are especially punitive and says that customers have generally not objected, understanding that the system is a privilege. Speaking on behalf of himself and others in the smart-cart industry, he quipped, "The customers don't mind these random checks nearly as much as we insist they do."

Both Cuesol's Grimes and NCR's Lutz agreed that RFID (radio frequency identification) is going to solve most of the self-checkout issues—and that it's not likely to happen anytime soon.

"The Holy Grail for all of this will be full-item RFID tagging," Lutz said, adding that he believes full RFID product tagging is still 10 to 15 years away.

The smart carts also have some practical disadvantages compared with stationary self-checkout systems. The smart-cart scanners require that the customer have a courtesy card, which means it couldn't be used when visiting another city or even a local store that wasn't a customer's main store.

In the pre-RFID world, Grimes said, Cuesol's cart tablets do a pretty good job of knowing precisely where products are, even when employees move product locations without updating the database.

Although the system can't yet zero in on one item, it uses a self-healing technology approach that realizes that customers are suddenly scanning cat food from the middle of aisle seven instead of aisle six, which is where the cat food is supposed to be. If enough customers do that, it concludes that pet foods have been moved to aisle seven and immediately updates its map accordingly, Grimes said.

Cuesol is also readying a system that tracks the actual carts, even if they don't happen to have a Cuesol tablet inside them. That way, the system could automatically flag management when, for example, a particular checkout lane had more than four customers in line.