Understanding Accelerometer & Gyro Drifts

Thread Starter


Joined Jul 13, 2014
Hi Guys,

I have been trying to understand the accelerometers and gyroscopes and have a few questions:

- How does integration produce the offsets in the readings when trying to calculate the position?
- Is the drift in accelerometers going to be more as compared to a gyro as we will integrate the acceleration twice?
- How can the accelerometer give you absolute orientation in the up-down plane?
- How can sensor fusion of accelerometer and gyro reduce the drifts?
- Lastly, if anyone has worked with mpu6050 in dmp mode does it already have algorithms to cater for the drifts?

Plus I actually wan't to measure angles produced by the hand rotations.

Thanks in advance.


Joined Feb 24, 2006
A change in position with respect to time is called velocity. A change in velocity with respect to time is called acceleration. Calculating the change in something with respect to time is called differentiation, or taking the derivative. Integration is the inverse operation of taking the derivative. In other words, we integrate acceleration to get velocity, and we integrate velocity to get position.

To understand why a gyroscope drifts, you need to work out all of the components of the forces on a rotating mass. In particular the interplay of small motions in one direction causing forces in a perpendicular direction. You know you understand vector algebra if you can follow this particular development. I'm less familiar with accelerometers so I can't speak for their behavior.

I'm not sure what you mean by absolute orientation in the up-down plane. In aircraft the location of position on the z-axis is set with respect to barometric pressure. In GPS that position is derived from phase differences in the arrival of the signals from the satellite. Not sure how you do it with just gyroscopes and accelerometers.