This summer at Bing we had a ‘hackday’ event. I worked together with Ravi Srinivas Ranganathan and Gabriella Ponce to make a Kinect and Bing Maps mash-up. Here’s a video of it being used.
KNUI Maps Demo from Ravi Srinivas Ranganathan on Vimeo.
The work was done in C# and Javascript, and used the Kinect SDK.
Supported operations include: * Pan left, right, down, up * Zoom in, out * toggle gesture detection * Voice detection to jump to predetermined locations/landmarks
Given that it was made for a hackday, it’s clearly a pretty rough project. Some simple and obvious extensions of the project would be to use an online voice transcription API to allow verbal searches for arbitrary locations (it currently uses local transcription, which is limited to detecting hardcoded phrases in this simple implementation), and to allow panning and zooming based on the activation of visible zones through hand hovering (instead of the full body gestures you see being performed in the video).
Still… with just those simple fixes, this would make a decent tool for world-exploration for a couch dweller.
And with some rather small advances in the underlying technology, mostly in processing speed and skeletal fidelity, we’d easily be able to make a ‘Minority Report’ style world browsing product. (And without those pesky gloves.)