Michael Ungar demonstrates a Microsoft wearable that uses facial recognition and voice commands to guide warehouse workers to items. The device learns each person's speech patterns, shows pictures of items at the bin, and catches picking errors instantly with two-finger taps to scan barcodes.
The beauty about the system we developed at Microsoft is the more the users use it, the system learns their speech impediments and tries to understand what they're trying to say. And it does voice-to-text and text-to-voice on screens that they can see, so to make sure that the machine is understanding what they're saying. But again, the more they use it, the more data the machine has to better understand what they're trying to say. The machine gives them turn-by-turn navigations in the warehouse. We wanted to do it with minimal training, so they could just log in with their face. They could say, I want to go pick. And it will give them turn-by-turn navigations on the screen and voice directions. You know, walk down this way. Okay, now turn left. You're approaching the bin on the right. Gives them turn-by-turn navigations. And then when they get to the bin they're supposed to pick from, shows them a picture of the item on the screen, and it shows them the bin and the barcode. They can scan it with just 2 clicks of their finger with the wearable. And if they scan the wrong bin, it'll say, hey, no, that's the wrong bin. You need to scan the one to the left or to the right.
