I am doing a project involving motion tracking using an accelerometer. I know a lot of approaches for motion tracking also involve a gyroscope, but here there is only an accelerometer available. In principle, the rotational information should also be obtainable by the linear accelerometer measurements, although the measurement error might corrupt this.
What algorithms are there specific to motion tracking using only an accelerometer? A general approach that I know is to use a Kalman filter for integration, which also attacks the integration drift problem. Is there a more specific approach for motion tracking using accelerometer data that is popular?
Should I try to recover the rotational information from the linear measurements?