One of the Carnegie Mellon University bookstore's newest hires isn't a bright-eyed college student, but a waist-high robot that looks a bit like a sweatshirt on wheels. The robot checks on inventory in store aisles, creating a map of stock that's sent to a touch-screen system in the store. The robot's inventor, Carnegie Mellon computer scientist Priya Narasimhan, thinks the bot is a better alternative to digital tags that many stores are considering for the future, MIT's Technology Review reported.
Narasimhan hopes the robot will help stores save money. She interviewed retailers and found that stores lose sales when they run out of popular items, and when staff don't know where items are and can't help customers find things. She'll get data on whether her robot improved the bookstore's bottom line in the fall.
The robot, which employees dressed in an orange Carnegie Mellon sweatshirt, uses a combination of computer science techniques to do its job. It is able to recognize objects it sees and learn from its past performance. It has a database of 3D and 2D images of store stock in its memory, as well as a map of where different objects, such as sweatshirts and mugs, are supposed to be displayed.
When it sees an object, it scans for the object's barcode, shape, size, color and any text on the object. It can also do a little elementary reasoning, based on the map in its memory. "If an unidentified bright orange box is near Clorox bleach, it will infer that the box is Tide detergent," Narasimhan told Technology Review.
Many researchers are now working on automated inventory systems like Narasimhan's robot, another robotics researcher, Ruzena Bajcsy at the University of California, Berkeley, told the magazine. Perhaps that means more people will get to see robotic store workers in the coming years. That's certainly true for shoppers in downtown New York City, where Carnegie Mellon is located. Narasimhan will test her robot in local stores next year.
Source: Technology Review