What is the best depth camera to be controlled from an fpga? intel realsense vs kinect v1 vs kinect v2?

Question

intel realsense vs kinect v1 vs kinect v2? What are the pro and cons of each of these sensors and what would be the best sensor to be used for an fpga implementation? where can I find datasheets or low level protocol information about these sensors?

Answer 1

The Kinect V1 is arguably obsolete. A few differences:

1. The Kinect V2 sensor is higher resolution, at 1920 x 1080 vs. 640 x 480 for Kinect V1.

2. Kinect V2 supports USB 3.0, so you can get a 10x improvement in the data transfer rate

3. Microsoft's new SDK2 is not backward compatible with Kinect V1. So if you want to work with the new SDK, you are locked into Kinect V2.

The easiest way to interface it to an FPGA would be to use a Windows system with an FPGA card. Microsoft sells the Kinect with the hopes that you would be writing an XBOX or Windows application, so you can find lots of support for those platforms.

From your question, I suspect you would like to design your own hardware. In this case, you can find information about the low level USB drivers at the OpenKinect GitHub site.

Note that whatever system you design, it will probably be easier to do if it has a CPU in it. It would be theoretically possible to implement the functionality provided by the driver code into an FPGA....but very painful. You can get an FPGA with an embedded Cortex A9 that runs Linux, like the ZYNC-7000, for around $60.