The Kinect Accelerator – Microsoft is now backing Kinect-based start up companies…
If you are a developer or existing team/startup focused on building a business that takes advantage of the Kinect and Natural User Interface technologies, then the Kinect Accelerator is where you need to be. Through this program, Microsoft is supporting entrepreneurs, engineers and innovators like you to bring to life a wide range of business ideas that leverage the limitless possibilities Kinect enables.
Original post by adafruit http://www.adafruit.com/blog/2011/11/20/the-kinect-accelerator-microsoft-is-now-backing-kinect-based-start-up-companies/
Comments
It looks like the next kinect might not use usb http://www.adafruit.com/blog/2011/11/28/next-generation-kinect-may-...
Over at MIT they've developed a quadcopter that uses the Kinect for visual odometry. Their copter uses a "flight computer, developed by the Pixhawk project at ETH Zurich [24], is a 1.86 GHz Core2Duo processor with GB of RAM. The computer is powerful enough to allow all of the real-time estimation and control algorithms to run onboard the vehicle." though they use offboard computing for mapping and drift correction. Here's a link: http://groups.csail.mit.edu/rrg/index.php?n=Main.VisualOdometryForG...
I didn't know about this issue but what springs to mind is that the kinect projects a parttern of spots with a IR light (a low-powered laser?) and then reads them back in with a camera to determine distance and perhaps there would just be too much interference outdoors or the dots would no longer be visible?
thanks, that is a shame. do you have any information about why it doesn't work outdoors?
The main problem with using kinect for arducopter is that it will not work outdoors, but will definitely allow for some cool arducopter indoor projects where GPS is not available.
@healthy,
I think you're right there...an ARM based Arduino could likely handle the mass of data from kinect.
@fab,
Yes, it would be a nice step after the optical flow stuff...more vision based robotics. I was thinking obstacle avoidance or even a visual replacement for the sonar!
It's the high-speed USB for the depth camera that's got me stuck at the moment.
It's just Microsoft trying to market their Kinect system. Read in the FAQ you may actually have to relocate to Seattle for the duration of the project. They don't mention who's paying for your room and board for that period. I suspect $20,000 in Seattle will not go very far.
based on this link we might have a long road ahead http://www.adafruit.com/blog/2011/11/21/yaskawa-electrics-next-gene... but using it for ground station waypoint and control might be interesting, but i guess randy is thinking optical flow :)
Would you be able to run the code you need on a 32-bit microcontroller? Maybe that's the next hardware iteration to allow running stuff like a Kinect?
Interesting..
Well, whether Microsoft coughs up 20k or not, I'd love to interface a kinect with arducopter (or other low powered microcontroller). I've managed to control the little light on the front of a kinect using the Sparkfun USB Host and the code written by "Oleg" of circuits at home... but to make use of the depth camera, the challenges appear to be:
Also by the way, the gesture motion code that you sometimes see all runs on the client (i.e. your PC or xbox). That's not something that could be easily put on a microcontroller!
If anyone's interested in sharing ideas on communicating with the kinect please ping me (or just post here i guess!).