"We have not deployed the cart yet in an airport, but we have deployed it for an extended time in a grocery store," said Utah State University professor Vladimir Kulyukin, who designed the cart together with four graduate students. One of those students, he said, "is an EE, and we would not have been able to accomplish our goal without that person."

The robotic cart first presents users with a Braille menu of items. When the shopper chooses a product, the cart leads the individual to the proper aisle and provides verbal, step-by-step directions on how to retrieve the desired item from the shelf.

The robotic shopping cart was built for less than $20,000 using off-the-shelf components that were integrated by the team's EE. Custom software was designed by the other team members for an embedded laptop computer connected to the robot, an RFID antenna, a Braille haptic display (reader) and a keypad input device.

The current design is the second generation. The original prototype attempted to use computer vision to perform pattern recognition of product boxes on the grocery shelf, but "the error rate was so high as to make it practically impossible with the current state of pattern-recognition algorithms," said Kulyukin.

Instead, the robotic cart now catalogs all the RFID tags in its vicinity, then organizes them into an internal road map. Users can scan through a catalog of the items on the road map using the Braille reader. When users find an item they would like to buy, they feel a number written in Braille next to it. A keypad on the back of the robot allows the user to key in that product number.

Once the robot knows which product the user desires, it navigates to the appropriate aisle. It then switches to verbal mode. For example, "if you key the code for toothpaste, when the cart arrives at the toothpaste aisle it announces, 'The toothpaste is on the third shelf on the right', " said Kulyukin.

Kulyukin is working to add a glove with a built-in bar code reader to the smart cart that would let users fine-tune their selections — among different brands of toothpaste, for example — by picking up the boxes and scanning their bar codes. The voice synthesizer on the robot would announce the name and price of the item.

The robot is not intended to replace a guide dog. Rather, the cart could take the place of the human companions on which the visually impaired must often rely for assistance when shopping or traveling. The cart would thus empower its users to do more on their own.

"I could go to a grocery store by myself if something like this were available to me," said Sachin Pavithran, a visually impaired test subject for the project. "It would help in so many places where I can't go alone now. For instance, when I am in an airport and have a flight layover, I am often stuck in one place because I can't get around by myself. This robot would give me back some independence."

Kulyukin also has plans for a wearable navigation system for the visually impaired that could direct them in outdoor environments, such as parks, that might not be familiar to their guide dog.

"I wanted to build something that actually makes a difference," said Kulyukin. "We are helping people, and nothing is more satisfying than that."

Utah State University, the Center for Persons with Disabilities, the National Science Foundation and the Community/University Research Initiative all helped fund the project.