The short figure creeping around the Carnegie Mellon University
campus store in a hooded sweatshirt recently isn't some shoplifter, but a robot
taking inventory. Andyvision, as it's called, scans the shelves to generate a
real-time interactive map of the store, which customers can browse via an
in-store screen. At the same time, the robot performs a detailed inventory
check, identifying each item on the shelves, and alerting employees if stock is
low or if an item has been misplaced. While making its rounds, the robot uses a
combination of image-processing and machine-learning algorithms; a database of
3D and 2D images showing the store's stock; and a basic map of the store's
layout—for example, where the T-shirts are stacked, and where the mugs live.
The robot has proximity sensors so that it
doesn't run into anything. None of the technologies it uses are new in
themselves. It's the combination of different types of algorithms running on a
low-power system that makes the system unique. The map generated by the robot
is sent to a large touch-screen system in the store and a real-time inventory
list is sent to iPad-carrying staff. The robot uses a few different tricks to
identify items. It looks for barcodes and text; and uses information about the
shape, size, and color of an object to determine its identity. These are all
pretty conventional computer-vision tasks. But the robot also identifies
objects based on information about the structure of the store, and items belong
next to each other.
More information: