Double integration of acceleration from accelerometer to find position?

Question

I am attempting to use output from an accelerometer moving back and forth on a single axis to calculate its current position.

I have tried using Euler integrations, but the velocity and position errors become too large too quickly. After some reading, I am wondering whether RK4 could be applied to this problem to minimise the error?

Cheers,

A.

Answer 1

RK4 might help.

Are you also measuring position, and that's where your assessment of accumulated error comes from?

It might also be due to time increments being too large. Perhaps you've exceeded a stability limit for your system. A finer time step and an implicit integration method might help with those errors.

Answer 2

There is usually some drift in accelerometers. You need a measurement of the actual position to remove static errors.