Introducing Pixy (aka CMUcam5). Its an embedded camera that has a dual core ARM processor, USB/I2C/UART/SPI communication, and built-in vision algorithms. I started working on this camera back in September 2012 as a graduate student at Carnegie Mellon, working with Rich LeGrand of Charmed Labs and Anthony Rowe, a CMU professor who created the first CMUcam as an undergrad (or so I hear). A Kickstarter for Pixy was just launched this morning with hopes that we could get enough backers to make a large production run in order to get manufacturing costs down, and thus get the camera in more people's hands.
So what can you do with Pixy? Probably the best feature is its color trainer and tracker. Simply place a colored object in front of the camera, hold the button down, and let go when the color of the LED matches that of the colored object. Doing this will train the camera to track any object of that color. This simplicity makes the camera extremely easy to use and perfect for just about any tracking application. Then using one of the many communication mediums you can interface with just about anything and control it based on what the camera finds for you.
Aside from the fact that I helped develop this device, I'm terrible at really conveying its features and usefulness. So be sure to watch the video above and check out our Kickstarter page!
UPDATE: I've created a new page dedicated to Pixy, which is where I'll be posting all future updates.