Hello Dirk,
Thank you for your post. One key technique in these types of applications is to use gyroscope-derived angle estimates to remove the gravity component from the accelerometers. Since the ADIS16305 does not have pitch and roll gyroscopes, it will not support this level of estimation and feedback for key linear state estimation (position, velocity, acceleration). Therefore, we would imagine that the ADIS16448 or ADIS16485 might be better to start with. Aside from the sensor function considerations, MEMS sensors seem to be reaching a point where their performance can contribute to the position tracking process, but not support indefinite position tracking on their own. The state of the technology, along with the many applications that would value this function have fueled a lot of activity; in other words, this is a hot field of research and development. As it stands today, the solution space seems to include a combination of MU-derived states with other observations, such as wheel-based odometers or relation positioning (radar, lidar, sonar, etc), using application-targeted algorithms.
I hope that helps!
Mark