Just found this website that includes a methodology to produce very accurate indoor navigation. They combine a technique called Rolling Shutter with IMU sensor fusion.

http://www.ee.ucr.edu/~mli/RollingShutterVIO.html

Here's the paper covering her methodology:
http://www.ee.ucr.edu/~mourikis/papers/Li2013ICRA.pdf

Don't be intimidated by the quaternion math because Invensense solves them for us in the MPU 6050.

I haven't yet found the source code but will post it here when I do. Here's some source code from Hannes Ovren who is integrating Rolling Shutter with gyro data:
github.com/hovren/crisp

testpypi.python.org/pypi/crisp/2.1rc11


Cheers,

Bill