I have no experience in Kalman filters and I'm afraid my maths is not
really up to what I want to do.

I'd like to implement a Attitude-Heading-Reference-Systems with a
small microcontroller sampling 3 accelerometers, 3 gyroscopes and 3
magnetometers and outputting the estimation for heading, yaw and
pitch.

I've been trying to read up on Kalman but am at a loss for convariance
matrices and other variables. So I would appreciate if I can get
explicit help on the code I need to implement.

Basically, the state vector should probably be [angle_yaw,
angle_pitch, angle_heading, acceleration_x, acceleration_y,
acceleration_z]. Or at least, that is the ouput I'd like to see from the system.

The measurements I will be taking are [accel_x, accel_y, accel_z,
rotation_x, rotation_y, rotation_z, magnetic_x, magnetic_y,
magnetic_z].

The application for the system would be to estimate the orientation of
a small model airplane to be able to display a artificial horizon and
magnetic heading to the pilot.

--Thomas

It's been awhile since I've attempted something like this.  Let me ask
1 question about your system, to determine the ease of implementation.
 Is is a linear system?  I'm betting it's not, since you're dealing
with rotation angles.  Just had to ask to start with.

This is really where it shows how little I know about the things I'm
asking about...
Could you explain what you mean by 'system'? My guess is that I need
to find a model for how the different variables relate to each other,
or? Can you give me some help on how to describe 'the system'?