Robots in Retail

Researchers at Carnegie Mellon University see the future of retail — with AndyVision, technology that helps to improve the shopping experience for both customers and retail staff.

The project is part of Priya Narasimhan's work as head of the Intel Science and Technology Center for Embedded Computing at CMU.

AndyVision — a robotic inventory system — takes the form of an autonomous robot that can patrol and scan the aisles and shelves of a retail store.

The robot generates a detailed aisle/shelf-level interactive store map that can be displayed on an in-store digital sign for customers to browse the virtual world of the store using gestures or a touchscreen interface.

The robot — a research prototype under exploration at the Carnegie Mellon University Store — can perform a thorough inventory and can tell store staff when an item is running low in stock or if merchandise is out of place.

AndyVision accomplishes all of this through the real-time fusion of image-processing and machine-learning algorithms, a database of images of the store's products, a basic map of the store's layout, and sensors to help the robot navigate autonomously and to prevent it from running into things.

"The idea for AndyVision was born out of me being a shopper. I go to a lot of stores and I find it very difficult to find the items I want, and sometimes I leave when I don't find what I want," said Narasimhan.

"As I talked to retailers, I realized that store staff don't always know where all of the items are located either. I saw a need for real-time product location and inventory information in everyone's hands. And what we are fundamentally trying to do is build technologies that put that kind of real-time product info into shoppers' hands as well as store staff's hands."

"We were thrilled to collaborate closely with the Carnegie Mellon University Store to develop, deploy and study these technologies in a real-world retail environment."

Ryan Wolfe, director of campus services at CMU, says retail staff face the traditional challenges such as inventory management, customer service, staying on top of or ahead of consumer trends, and of course watching that bottom line.

"I believe AndyVision can make a real difference in retail," Wolfe said. "The potential labor savings associated with robotically-controlled inventory management as well as the potential impact it can have on the distribution of accurate, store-specific planograms (a visual representation of a store's products/services), are both exciting. And it has the ability to integrate some of the benefits of a virtual shopping experience into a brick and mortar store such as improved navigation and fast product location."

Intel Chief Technology Officer Justin Rattner said, "The AndyVision project represents the cutting edge of embedded systems research using a combination of advanced robotics and computer vision. If successful, the project will make future retail stores more successful while ensuring customers find the products they want to buy."

Rattner added, "This is exactly the type of game-changing research we expect from the Intel Science and Technology Center at CMU."

AndyVision team members include CMU Information Networking Institute/Electrical and Computer Engineering (ECE) Senior Systems Scientist Rajeev Gandhi; CMU ECE post-doctoral researcher Kunal Mankodiya; CMU ECE students Jonathan Francis, Amit Krishnan and Utsav Drolia; CMU Director of Campus Services Ryan Wolfe; and Angela Tumolo Neira, marketing specialist at the Carnegie Mellon University Store.