IMU - a pointer in the right direction?

[javalin]

yeah a good link.

I've got your Quanternion's implemenation working in vb.net with graph's and your Artifical horizon control (tweaked) - and thats nice, but it drifts like a bugger.

Onto Mr Kalman.

James

[Tom]

If it drifts you need to tweak the constants a bit

[javalin]

interesting - which constants? Also I haven't implemented the PID from your code - just the quanternions.

James

[Tom]

Oh ok,
So you took the code from ahrs_quaternion... You need to implemented the PID-part as well to compensate for drift. This is a simple complementary filter.

For your application, euler angles will work just fine!

[javalin]

yes - so a Euler angle would be just X, Y and Z as degree's -90 -> +90 wouldn't it? Just simple addition and subtraction to modify, and trig to generate?

This would work for what you posted (brackets added for me):
Position = Last_Position + (Speed * dt)
Speed = Last_Speed + (Acceleration * dt)
Roll_Angle = Last_Roll_Angle + (Roll_Angle_Rate * dt)

and I can do pitch the same?
Pitch_angle = Last_Pitch_Angle + (Roll_Pitch_Rate * dt)

And the roll_angle_rate is the deg/s from the gyro I assume.

I assume this then forms the mathmatical model which I can then work on drift,etc, correction?

Thanks!

James

[Tom]

Yes,

Keep in mind that this is simplified and will only work for small pitch and roll angles (which is true in your case).

[javalin]

HIya

Yes - apart from yaw I suppose, but the size of the changes between readings will be small if thats what you mean.

James

[Tom]

Nop,

(Do this with your hands)

The roll angle is 60 degrees. Pitch 40° up (as your pitch gyro would understand). You'll see that your true pitch angle is not 40° but ~ 40° * cos(60°)= 20°