We were able to successfully combine the hand tracking from the Kinect with the 3 cubes we had constructed earlier. The tricky part of this integration was deciding on a protocol to communicate between the Raspberry Pi and the mbed. We ended up encoding the values into chars and sending that data over serial.
The other challenge was figuring out what cube should actuate based on hand position. We decided to split up the Kinect's depth camera's resolution into a grid based on the number of cubes we had (in our case 3x1). The algorithm from there was to determine which grid cell the user's hand was in (using the calculated centroid) and actuate the corresponding cube.
Here's a video of that working:
https://goo.gl/photos/t1Zf1Td3pkmxRz9JA
No comments:
Post a Comment