0
\$\begingroup\$

I'm trying to build a piece of hardware with an accelerometer that could track the approximate 3D position of an object. The accelerometer would not be rotating, so a gyroscope shouldn't need to be accounted for. I would like to know if anyone has any suggestions for a specific accelerometer. Does anybody have suggestions for a quality accelerometer?

Also, does the project sound doable? Could an accelerometer accurately detect the position of an object? This would all be on a small scale; specifically the tracking of one's finger. A GPS wouldn't work in this situation, as the movement would only be moved about in around a 2'x2'x2' field.

Thanks.

\$\endgroup\$

closed as off-topic by Leon Heller, Chetan Bhargava, Matt Young, Joe Hass, Stephen Collings Jan 25 '14 at 23:54

This question appears to be off-topic. The users who voted to close gave this specific reason:

If this question can be reworded to fit the rules in the help center, please edit the question.

  • 2
    \$\begingroup\$ You might like to search for 'inertial dead reckoning' and read some of the articles that are around on why this is very difficult. \$\endgroup\$ – David Jan 25 '14 at 21:54
3
\$\begingroup\$

To get a position from pure accelerometer data, you need to integrate it twice over time. First integral gives you velocity, second gives you position.

The problem you run into is called integration error. This error gets larger and larger over time. And it gets much worse when you're integrating twice.

And what's more, for integration, you need input conditions, that is - you need to know position and velocity of the object at the moment you start the measurement.

So in the end, you may be able to track the direction of movement, you could measure the position during the first second of time or so. But it will be far from accurate, and you will need to assume you're starting from a stand still.


I suggest looking at positional measurement, rather than acceleration. Such as optical, magnetic, sound, etc.

\$\endgroup\$
  • \$\begingroup\$ Do you have any specific recommendations in general? Like a piece of hardware? \$\endgroup\$ – mjkaufer Jan 25 '14 at 21:20
1
\$\begingroup\$

To measure changes in position in general you need three gyroscopes and three accelerometers (6 DOF). If you can fuse data from other sources (compass, GPS, triangulation) you can get low drift and fast response. The optimal way to fuse such data is usually a Kalman filter. In your case, maybe triangulation would be practical, augmented by MEMS accelerometers and perhaps gyros.

Really good accelerometers (low bias drift, linear over a wide range, low temperature drift) are very expensive and somewhat bulky. Anything suitable for a tactical or navigation grade Inertial Measurement Unit (IMU) will be many thousands of dollars (and usually export-controlled, which causes all sorts of headaches).

\$\endgroup\$
0
\$\begingroup\$

With some math (Forward and Inverse kinematics) and more than one accelerometer (preferably one for each joint of the finger), you can track a finger. you will have to translate the acceleration data into orientation relative to earth's center, then translate that into relative angles (finger relative to the hand), and calibrate for the user's hand size. You will be able to track only the finger, not the entire hand.

\$\endgroup\$

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