Jump to content

Recommended Posts

Dear all,

Can anybody give me some links/references to good tutorials (introductory tutorials) to Kalman filtering?

Also can anybody explain me if the infrared horizon detecting system would work above water? (above a sea, ocean)?

I've noticed some Inertial Measurement Units (IMUs) on the market and their producers say that those units are drift free. How can a gyro&accelerometer unit be drift free when we all know that gyros and accelerometers have all kinds of drift...? Does that mean that I can fly their IMUs for 4 hours and not have any drift at all?

Thanks,

Soyuz

Share this post


Link to post
Share on other sites

Because they also have magnetometers, that have a reference.

Accelerometers will allow you to correct drift on gyros pretty well, but only on pitch / roll, not yaw. If you need that or also want good dynamic performance (like in flight where the acceleration isn't only static), you have to throw in the magnetometers.

Share this post


Link to post
Share on other sites
Because they also have magnetometers, that have a reference.

Accelerometers will allow you to correct drift on gyros pretty well, but only on pitch / roll, not yaw. If you need that or also want good dynamic performance (like in flight where the acceleration isn't only static), you have to throw in the magnetometers.

Hi there!

Thank you for your answer.

I read about using magnetometers in order to correct the "attitude readings". My understanding is that these magnetometers get a reference with respect to the local Earth magnetic field... is this correct? How do you know where the local earth magnetic field points? I assume one could find this information from some maps or somewhere online?

Even if one can find this information, what do you do when you fly near magnetic anomalies? I mean what if there is iron accumulation underground.. wouldn't that throw off the magnetometer readings? And if this is the case, wouldn't the IMU stop working correctly in this case?

Thanks,

Soyuz

Share this post


Link to post
Share on other sites

I did some more thinking on the magnetometer problem... What about using a tri-axis magnetometer to get the orientation in space of the airplane and then substract from the 3 accelerations (read by a 3 axis accelerometer aligned with the magnetometer) read by the accelerometer the corresponding g-components?

That would leave us with ax,ay,az accelerations of the airplane along its fixed coordinate system...

That should be enough to get the total attitude of the airplane, shouldn't it?

As far as I see (if I did not do any mistakes) this system shouldn't need gyros...

Thanks,

Soyuz

Share this post


Link to post
Share on other sites

Soyuz (are you Russian?)

Check Sparkfun (http://www.sparkfun.com) their 5DOF module (five degrees of freedom) has all you need basically, not you just need to interface it with a MCU and write some code to make it all work.

Cheers,

Sander.

Share this post


Link to post
Share on other sites
Soyuz (are you Russian?)

Check Sparkfun (http://www.sparkfun.com) their 5DOF module (five degrees of freedom) has all you need basically, not you just need to interface it with a MCU and write some code to make it all work.

Cheers,

Sander.

I would need complete 6 degrees of freedom... as far as I see the 6 dof are quite expensive event sparkfun's ones...

Would the magnetometer+accelerometer idea work? Has anybody tried it before?

I don't see why it would not work...

Soyuz

P.S.: I'm not russian but I like Soyuz capsule :)

Share this post


Link to post
Share on other sites

Soyuz,

I you want to build something that stabilizes the aircraft's framel like Carvec does, you might wanna read here:

http://en.wikipedia.org/wiki/Inertial_guidance

this is actually it's principle of operation.

There is an article in Wikipedia about Kalman filtering too (I got a headacke trying to read it - not to mention that I didn't understand a thing :-( )

Dimitris

Share this post


Link to post
Share on other sites
What about using a tri-axis magnetometer to get the orientation in space of the airplane

The problem is that magnetometers are fairly slow and inaccurate.

A full IMU uses gyros, accelerometers and magnetometers as each of them has advantages, but will also be thrown off in certain conditions. As the conditions are different, there's a way to tell when one is going off as at least one of the others won't.

For example if a current spike is throwing your magnetometer off for 1 second, and it suddenly shows a 90° rotation, but the gyro hasn't detected any movement, you know the magneto is giving wrong info and discard it, etc.

Kalman filtering is just a mathematical way of isolating the real info that is reported coherently by the sensors while discarding inconsistent data.

Share this post


Link to post
Share on other sites

The problem is that magnetometers are fairly slow and inaccurate.

A full IMU uses gyros, accelerometers and magnetometers as each of them has advantages, but will also be thrown off in certain conditions. As the conditions are different, there's a way to tell when one is going off as at least one of the others won't.

For example if a current spike is throwing your magnetometer off for 1 second, and it suddenly shows a 90° rotation, but the gyro hasn't detected any movement, you know the magneto is giving wrong info and discard it, etc.

Kalman filtering is just a mathematical way of isolating the real info that is reported coherently by the sensors while discarding inconsistent data.

That makes things WAY clearer! Thank you very much Kilrah!

I was wondering why all the INU manufacturers use accelerometers, gyroscopes and magnetometers and sometimes even a GPS... now this is clearer to me!

I am now investigating the possibility of building a Inertial Measurement Unit enhanced with an infrared horizon sensing device.

Are the infrared sensors also slow (like the magnetometers)? I ask you these questions because I want to eliminate the possibility of going on a wrong route in this design... people that have experience in sensors can immediatly tell things/give advices like Kilrah's one!

Thank you again!

Soyuz

Share this post


Link to post
Share on other sites

Hello Soyuz,

Kalman filtering is just an estimation technique. What's more important, is what you are estimating. In most cases it is the state of a mathematical model representing the aircraft kinetics. You have sensors but they are (1) too slow or (2) can have a lof of noise/drift. With a kalman filter you can "fuse" them.

Difficult things to estimate are: centripetal accelerations, accelerations on any of the 3 axis.

You can find more information on my blog (i still need to cover a lot of topics ;-) )

http://tom.pycke.be/mav/71/kalman-filtering-of-imu-data

Tom

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×