Greetings ADI MEMS Team,
We're using the ADXL345 (your development board logger) for an exciting sports analytics application.
This involves a relatively unstructured environment...meaning its not a gym floor or field where gravity is perpendicular to the floor. In our case, the logger is attached to the subject, and the subject is moving and their orientation wrt to gravity is highly changeable. The gravity vector moves throughout all 3 axes to some degree. The subject is moving with changing velocity causing accelerations, but again, throughout multiple axes.
Question is how to analyze the data to glean the "aggregate acceleration" at any given time?
Are there analysis techniques that people use to handle cases of acceleration in all 3 axes?
Also interested in digital filtering options. We're currently applying moving average techniques which help clear up the data. Are there more advanced techniques? The level of analysis is currently purely time-domain based. Not really looking at frequency response yet, although we may be interested in doing this on key segments of the data in the future.