From reading the article, there is no reason why Android was chosen specifically. So this experiment could have been done with any modern smartphone out there, they just happen to choose a couple of android phones. Is that right?
I've got nothing to do with this project and only know what I read in the linked article but the seemingly obvious reason they'd use Android phones, especially Nexus phones, is that they run on a nearly fully open source stack (minus some binary drivers), so you can go as deep into the internals as you want without resorting to jailbreaking and/or a lot of reverse engineering.
Other NASA pages about this project seem to emphasize the "Open source" qualities of the project. There's also language like "with the goal of allowing anyone with space ambitions to launch their own satellite".
For instance, most other major options (elephant in the room being Apple), have onerous terms for their accessory development kits which would preclude that. Last I checked, Apple even keep the Dock port pinout "confidential" from people not in their accessory dev program, which would be a show-stopper since presumably this uses the dock/USB port for control.
My guess is the same reason ArduSat uses phone hardware.
All the sensors one might want on a Satellite (motions sensors, cameras, magnetic field detection) are all built into phones now. Android (or more to the point Android's Linux kernel) is specifically compiled to be compatible with most of that hardware. So less work for them as opposed to using another version of Linux.
At which point the satellites can no longer be open source. Perhaps more importantly, Android already runs on the Nexus S; VxWorks doesn't. It defeats the purpose of using commodity hardware if you start over from scratch on the software running it.
I agree VxWorks would require additional porting effort. I also think that Android was the right choice for this mission.
I'm not sure I agree with the implied high priority placed on open source. The relative development costs of android, real-time linux, or even a commerical RTOS are noise in the flight software development budget. Reducing the cost is a more important driver to making space more accessible.
The big benefit, if these android cube sats fly successfully, would be to retire a lot of risks - thus opening the door for future programs to fly with android.
