1

This is strictly not aviation related so please feel free to move it to the correct section.

I have a 9 axis IMU, which gives me yaw, pitch, roll, magnetics, gyros, accels etc.

I currently make use of the roll, pitch and yaw (true) however I want to learn how to use the rest of the data. I know that the raw magnetometer data can be used to judge when the true yaw is not accurate enough and even correct it. I'm reading up more on that subject at the moment.

However what I can't find much resources on is on how to make use of linear acceleration and angular acceleration. I have successfully converted the data from raw units to g/s.

How do I obtain the acceleration of my craft with the data?

menjaraz
  • 2,435
  • 3
  • 18
  • 30
Nopestradamus
  • 11
  • 1

1 Answers1

2

You integrate accelerations to get speed and integrate speed to get position. So if you have angular speeds or even orientation, I don't see any use for angular accelerations. That does not mean the unit does not use them internally when calculating the angular speeds and/or orientation.

The linear accelerations can be used for dead reckoning. Airliners, and increasingly any plane with glass cockpit, have inertial navigation system that integrates the accelerometer and gyroscope data to keep track of position. It used to be primary means of navigation over ocean before GPS and now it is still used as backup if GPS signal is lost and for compensating errors by cross-referencing.

The lateral acceleration is also shown by the slip-skid indicator (used to be accompanied by turn rate in turn indicator, but glass cockpits often only have attitude indicator provided it is non-tumbling and maintains precision at unusual attitudes; see also When/How may I replace the Turn Coordinator with an Artificial Horizon?).

Community
  • 1
Jan Hudec
  • 56,247
  • 12
  • 155
  • 268
  • I don't think that normal discrete integration will work? Because of all the noise that will come into play? – Nopestradamus Jun 25 '14 at 08:51
  • 2
    @spaghetti-: It is digital, so it is necessarily discrete integration. Of course it's not just adding numbers. Numerics folks have various methods for filtering out noise (usually Kalman filters; may be already implemented in the unit) and interpolating smooth curve through the data points and integrating that to increase the precision (best known are probably Runge-Kutta methods). I haven't done much numerics myself so I don't know what exactly would be used or how to tune it. – Jan Hudec Jun 25 '14 at 09:21