Evan Schuman's StorefrontBacktalk

Wal-Mart on Thursday started quietly testing a university-created robot designed to help visually impaired consumers navigate store aisles and find their desired products.

The robot?named RG, for Robotic Guide?is the creation of Vladimir Kulyukin, an assistant professor of computer science at Utah State University and the director of the university’s Computer Science Assistive Technology Laboratory.

The initial version of RG?which weighs about 22 pounds and is roughly the height of an upright vacuum cleaner?is limited to three basic functions.

First, it guides the consumer through the aisles and around people, displays and merchandise using RFID readers and 16 ultrasonic sonars. The navigation system is sophisticated enough to handle environments?including elevators and limited open spaces?that usually literally trip up robots, Kulyukin said.

The university has posted quite a few videos of RG in various stages of testing.

Its second function is to communicate with the consumer. It takes instructions via a small Braille directory of products that is attached to the robot’s handle, and it replies to the shopper’s questions with spoken answers.

The third function is to use its RFID reader to locate the desired products. The store’s RFID tags help the robot navigate the lanes as well as locate products.

“There are RFID sensors placed on the shelves in the store. The robot has the RFID antennae and detects the presence of those tags,” Kulyukin said. “That’s how it knows it’s reached the Colgate section of the toothpaste shelf and it then announces, ‘You have reached the Colgate toothpaste section, on your right.'”

The robot has its limitations, though. Until item-level tagging becomes the norm, the system can indicate only the part of the shelf where the product is supposed to be. If it’s been moved?either by an employee moving stock who forgot to move the update the RFID tag or by another consumer who put a tube of Aim toothpaste amidst the Colgate?the visually impaired consumer might grab the wrong product.

“It certainly can be jumbled, and there is the potential to pick up the wrong product,” Kulyukin said, adding that his team is trying to add a robotic bar code into the system so that the robot would announce the product being placed in the cart. That functionality would likely address most of the mistaken product purchases, he said.

The robot’s development is still at a very early stage and has thus far mostly been paid for with a $500,000 grant from the National Science Foundation, Kulyukin said. He is negotiating with a large national retail chain to buy the units and invest in its further development.

Kulyukin refused to identify the chain, but an employee in the university’s public relations department, Whitney Wilkinson, said the chain was indeed Wal-Mart. Kulyukin also said Wal-Mart was testing it locally.

The store manager of the Wal-Mart store in North Logan, Utah, right near the university’s labs, confirmed that RG had arrived on Thursday.

“It’s a great thing for the customers who don’t have their eyesight,” said Wal-Mart store manager Ron Tuttle. “We have a lot of customers who come in and ask for someone to help them. I talked with one lady and she was very excited about it because it makes her feel more independent.”

The cost of the robots will vary depending on how many of them Kulyukin’s team is asked to create, but he purposely kept the cost low. To create a second robot would cost him about $10,000, he said, adding that the per-unit cost would drop to about $4,000 to $5,000 if thousands were ordered and to about $1,500 if millions were ordered. He said he will need about $3 million to $5 million in seed money to move to the next stage of development and production.

The technical hardware of the system is simple: Most of the components sit in a PVC pipe structure. The robot’s microcontroller is attached to a laptop, with which it communicates via serial cable. The laptop also can communicate using an 802.11b wireless card.

Kulyukin said he has spent much of his life focused on using technology to help those with physical challenges, partly to help his brother, who has always had severe hearing disabilities. “Growing up as the brother of a disabled child, I know firsthand how harsh the environment can be on you,” he said.

The problem of blind shoppers is fairly widespread. The National Institutes of Health’s National Eye Institute estimates that 80 million Americans today have potentially blinding eye diseases and 1.1 million people are legally blind.

“Approximately 12 million people have some degree of visual impairment that cannot be corrected by glasses, and more than 100 million people need corrective lenses to see properly,” the institute’s Web site says.

Given that RG is only in very limited experimentation today, how do most blind consumers shop? “They simply don’t go grocery shopping,” Kulyukin said. “If you happen to have a sighted spouse or a friend, that’s what you do. [RG] is an independence device.”

The business side of the retail argument for these robots is twofold. First, the people who shop for those blind consumers might not shop at the places those consumers would want. These kinds of robots would return the store-selection power back to those consumers.

Secondly, not many grocery stores have the financial resources of a Wal-Mart to invest in this level of robotic technology. Arguably, this could be a major differentiating factor in bringing visually impaired customers?and their friends and families?to Wal-Mart who might otherwise have shopped at the competition.

Kulyukin also said that having a small squadron of robots around a retail shop could be valuable in other ways. When there are no customers using the robots, they can assist in moving merchandise, carrying extremely heavy boxes and unloading trucks. After all, what good is having a bionic robotic arm if it’s not flexed once in a while?

The robot “doesn’t have to sit idly in the store. It can optimize the store’s supply operations,” Kulyukin said. “Instead of letting a truck come to the store and having it unloaded manually, load it onto the robot and then let the robot deliver it.”