I have x, y and z coordinates for a test I had carried out for a tool. I had an accelerometer installed on the tool and I used two types of motion to move the tool, rotation and translation. Right now I have x, y and z coordinates of the object as it moved and they are recorded with respect to a certain time interval of approximately 0.037s and I need to calculate its acceleration. do you know if this is possible? please let me know if you have any ideas.

  • $\begingroup$ Out of interest, why do you need to calculate the acceleration if you had an accelerometer installed on it? I think I must be picturing the situation wrong. $\endgroup$ – jacob1729 May 28 '19 at 12:49

I need to calculate its acceleration. do you know if this is possible?

No, it is probably not possible.

Since you allowed some rotation, and probably did not strictly rotate about the center of the accelerometer the rotating motion about a different point probably induced some additional signal to the accelerometers.

Imagine mounting an accelerometer to the edge of a wheel and spinning the wheel. Even if you don't translate the wheel's center, your sensor will still report a strong acceleration along the line which is normal to the rotation axis passing through the accelerometer.

If you want to do a better job, here are two things you might try:

  1. Only do careful rotations of your object around the center of your accelerometer. The problem here is that you don't know where it is exactly and the X, Y, and Z accelerometers are in slightly different positions.
  2. Use a six-axis device; three acceleration axes (proper accelerometer) and three rotation axes (gyroscope). Know the vector distance between the accelerometer and the gyroscope and then go through some math to predict and subtract the induced acceleration due to rotation to recover the linear acceleration.
  3. Mount two 3D accelerometers opposite each other, equidistant from the object's center, and *only rotate the object around this center.
  4. Read further for other strategies on how to use accelerometers to reconstruct trajectories.
  5. Don't rotate your object.

Good luck!

Learn a little about Kalman filters while reading further, that will be your next step.

  • 1
    $\begingroup$ Might it be worth adding how to deal with the translation-only case in this answer? I agree that without a lot more details about the rotation this seems hopeless. $\endgroup$ – jacob1729 May 28 '19 at 9:20
  • 1
    $\begingroup$ @jacob1729 I just added a link to Kalman filters, but I think that numerical estimators of higher order derivatives of a data stream of positions can be dealt with better on several other SE sites; it's not physics. Rigid body rotation concepts are on-topic in Physics SE, so I've limited my answer thusly. However if you posted a concise and helpful answer about that I would certainly up vote it! $\endgroup$ – uhoh May 28 '19 at 9:23
  • 1
    $\begingroup$ @uhoh thanks for the information, in fact this is my first time using an accelerometer and I am not familiar with it or what kind of data it yields but thanks for the help $\endgroup$ – Majid ALZADJALI May 29 '19 at 12:44
  • $\begingroup$ @MajidALZADJALI It's really fun stuff but there is a bit of a learning curve. You can also browse Q&A in Robotics SE and Raspberry Pi and Arduino SE and Engineering SE for existing posts on accelerometers and gyros. For (fairly advanced) signal processing also have a look in DSP SE. $\endgroup$ – uhoh May 29 '19 at 15:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.