I have been working on a new interface board for the Jetson TK1 embedded supercomputer called the Jetduino. It makes it very easy to build robots that can use the parallel processing capabilites of the NVIDIA GPU for vision and neural neural networks.
The Jetduino mounts above the Jetson and has a small connector that fits in the 2mm J3A connectors. A Raspberry Pi GPIO ribbon cable then connects it to the Jetduino. It has mounting points for a 2.5" HD, wireless antennas, a large proto-typing area, and a built-in shield for an Arduino Due. Just like the GrovePi for the Rapsberry Pi, you can use Python and C libraries to talk directly to the Arduino to set and receive digital and analog data. You can also control regular and smart servos. It has numerous Grove and RobotGeek connectors for modular sensors and motor actuators. I just put out a new blog post and two YouTube videos showing how to setup the Jetduino and perform digital I/O with 12 of the Jetson TK1 GPIO lines, and the 54 lines available on the Due. I am working on videos to show off the other features as well. If you would like to be notified when these are available please sign up for my newsletter. I plan to launch a crowdfunding campaign to get the Jetduino produced sometime in March or April, and I will need the help of any makers out there who want to make it easier to build robots or electronic projects with the awesome Jetson TK1.
Blog Post with YouTube videos: http://neurorobotictech.com/Community/Blog/tabid/184/ID/29/Jetduino-setup-and-digital-IO
Here are a couple of links describing what the Jetduino is.
Jetduino V1 description: http://neurorobotictech.com/Community/Blog/tabid/184/ID/25/Jetduino-prototype-boards-ordered
Jetduino V1 test results: http://neurorobotictech.com/Community/Blog/tabid/184/ID/26/Jetduino-V01-Test-Results