Underneath that color-coordinated hoodie is AndyVision, Carnegie Mellon's inventory assistance robot. It's programmed to take over the utter drudgery* of daily retail inventory, helping stores figure out what customers want and customers figure out how to get it.
You'll probably recognize the Kinect sensor underneath AndyVision's hoody, and he's also got a fairly simple mobile base with sonar for obstacle avoidance. Using Kinect, AndyVision scans store shelves to count items for inventory (using contextual object recognition), and will wirelessly alert store staff to low stock, no stock, or items that have been misplaced. Meanwhile, customers get access to real-time data on what items are where and how many are left. The technology involved isn't anything new and crazy, but this is a great example of a (relatively) simple robot being used to do valuable autonomous work in a commercial environment.
AndyVision is a project from the Intel Science and Technology Center (ISTC) at CMU, and he's part of its "Retail 2020" project to "transform the retail landscape." He's fairly retail-futuristic as is, but ISTC has bigger plans for the future, where "in-store robots might handle tasks such as folding clothing items, stocking shelves, and helping customers to locate items and load their purchases into their cars." Note that none of this stuff is really that futuristic, and it could be closer to "Retail 2015" than "Retail 2020." All it's going to take as an inexpensive and easy to use platform, some clever programming, and willingness on the retail side to try out something new.
*I speak from experience, having done retail inventory. It's not fun.