Best Sensor for Measuring Slow Head movement

Giray

Hi All,

I would like to hear your experiences and knowledge about measuring slow rotational movements by any kind of sensor. My goal is to measure moves of head accurately(up-down, left-right ) I think the best choices for this issue are accelometer and gyro I assume. I would be happy to hear your comments.


Regards,

Giray

Leon

When I worked on advanced cockpit research at BAe, we used one of these systems for measuring head and hand movement:

http://www.ascension-tech.com/

We had a very large budget, of course.

You won't get accurate results with an accelerometer and gyro.

ElectricAye

For a low-budget system, could you use a compass system plus perhaps a number of tilt sensors/accelerometers for up and down?

http://www.parallax.com/StoreSearchResults/tabid/768/txtSearch/compass/List/0/SortField/4/ProductID/779/Default.aspx

http://www.parallax.com/StoreSearchResults/tabid/768/txtSearch/tilt/List/0/SortField/4/ProductID/93/Default.aspx

Dave Hein

A gyro and accelerometer should work OK for your application. You will also need a compass unit to compensate for drift on the gyro. SparkFun has a 9 DOF unit that includes all three of these sensors.

ElectricAye

Dave Hein wrote: »

A gyro and accelerometer should work OK for your application. You will also need a compass unit to compensate for drift on the gyro. ....

If the movements are restricted to always being slow, is there really any need for a gyro?

Giray

Thanks for replies. My major concern is actually slow movements (in daily life rotations) whether they can be sensored adequately. Lets say error range is %3.

JasonDorie

It depends a lot on what you mean by "slow". The ITG-3200 gyro measures up to 2000 degrees per second, with 14.75 bits per degree-per-second accuracy. So, it could measure as low as 14th of a degree per second, relatively inaccurately, or 1 degree per second, accurate to about 1/15th of that, etc.. 1 degree per second is pretty slow for a head motion.

A 3-axis gyro, accelerometer, and compass would likely be necessary. The gyro will give you rotation rates, but is subject to drift from to temperature changes, numerical inaccuracy, sample quantization, etc. The accelerometer gives you "down", but is really noisy, so you have to average the results over a long-ish period of time. The compass will give you magnetic north. The three sensors together can complement each other - You use the accelerometer and compass to correct the long-term drift from the gyros, and you get a very accurate estimate of your orientation.

Here's an example done with just a gyro and accelerometer: http://www.youtube.com/watch?v=4bj2ieD4HA0

Pololu has a sensor called a MinIMU-9 that includes all three. I'm in the middle of writing a Propeller driver for it, but there are Arduino libraries for it already, and a demo program that shows the orientation of it on a PC: http://www.pololu.com/catalog/product/1265

You might be able to get away with just the accelerometer and gyro, but your response time for tilting left/right or forward/back would be pretty slow because the accel data is so noisy it requires pretty heavy filtering.

ElectricAye

Giray wrote: »

Thanks for replies. My major concern is actually slow movements (in daily life rotations) whether they can be sensored adequately. Lets say error range is %3.

For a human moving around in regular life, then I would guess you would need both a gyro and compass, as Dave Hein described, plus accelerometers. As for error, I guess that starts to depend on how long the device will be allowed to drift without getting a new absolute fix on something. Maybe if you describe what you're trying to do in more detail, people here can give you a better idea of the errors that might be involved.

EDIT: yeah, what Jason said.

Giray

Thanks Guys for useful replies. As a beginner for microcontroller and sensors your replies are very instructive.