How to export Openni2 on an embedded board (running Angstrom Linux)

Note: This question may or may not concern the OpenNI community directly as such, but I'm looking for any direction from anyone who might be familar with this. It might also be a bit broad, but I'll try to consider any feedback on this.

I have an industrial Texas Instruments embedded board with an ARM Cortex-A15, which is running an Arago Linux distribution by TI, forked from the Yocto Project (Angstrom type I believe). I wanted to get the Kinect v1 running on this board which means somehow I need the Openni2 drivers installed on this.

Note that this embedded board is not like your general-purpose Jetson TK1 or Beagleboard where you have a desktop version of ARM Linux running and you can sudo apt-get install packages and OpenNI2 for ARM and anything else you want. This is a bare-bones version of Linux where, to install new packages, a lot of times you have to complie the Linux code with package added, on an board-specific-SDK (Code Composer by TI) running on a host machine (Linux) and then port the code to the board via a bootable SD card.

I was wondering if there is any way for me to add these drivers as one blob of code without compiling or installing (either on the board or on the SDK on the host machine). For example, in Windows we can include Openni2 as a library (.DLLs etc) in our code, export this to the target machine and Kinect can access it on the target machine.

If there isn't a way to export Openni2 as a library, what steps would I need to make this happen? I Imagine if I copy ALL the files that Kinect needs from Openni and SensorKinect to the right locations this might work?

...Or, another way to ask this question is, is it possible to install Openni2 (either with the steps given modified, or by placing the right libraries in the right places) on an Angstrom Distribution?

Please let me know if you need more details.

OpenNI / OpenNI2

How to export Openni2 on an embedded board (running Angstrom Linux) #102