Robots could keep an eye on department store inventory

Robots could keep an eye on department store inventory

Like what you see? Tell a colleague
RSS Feed

News & Events - Engineering News

June 29, 2012

Shoppers at the Carnegie Mellon University (CMU) campus bookstore may see an R2D2-sized figure roving around in a red hooded sweatshirt, but they shouldn't be alarmed, the school says. It's just Andyvision, as the robot is called, and it is simply scanning shelves to create a real-time map of the store's inventory.

According to MIT's Technology Review, customers shopping in the store can view Andyvision's regularly updated maps to browse what is available and where it is located. The robot also collects information on stock and availability data and compiles it for employees to review.

The prototype, which came after extensive engineering research at the university's Intel Science and Technology Center in Embedded Computing (ISTCEC), was developed by Priya Narasimhan, a CMU professor. As the robot wheels through the store, it uses a combination of image-processing and machine-learning algorithms, a stored memory of images of what the store contains and a simple map of the store's layout. The machine was also outfitted with advanced proximity sensors to ensure it doesn't run into anything - including shoppers - around the store.

Narasimhan developed the robot to use several tricks to identify the items in the store, including searching for barcodes and text and graphing the shape, size and color of an object. While these are pretty typical features for a robot's object-identifying systems, it can also identify items based on information about the store's layout.

"If an unidentified bright orange box is near Clorox bleach, it will infer that the box is Tide detergent," she said.

The solution to use a robot to perform such activities could even beat out wireless RFID tags, as the computer-vision inventory system would be easier to implement than tags, the media outlet stated. Stores with metal shelves would have trouble installing the tags, especially if they needed to be placed on every item.

Ruzena Bajcsy, a University of California, Berkeley professor, said Carnegie Mellon isn't alone in developing such automated inventory systems. But for all parties interested in creating a similar device, the biggest challenge will be whether it "can deal with different illuminations and adapt to different environments."

ISTCEC is working on embedded computing systems due to their ability to be projected across various uses, including vehicle GPS receivers, HVAC systems and networked thermostats.

Read more like this