[Autopilot] Kalman filter questions & FAQ

[Electropiloto <ele...@ya...>]

Hello to everyone:

After searching and reading a lot of info in the net about Kalman filters
I'm now studying the code that corresponds to the Six state Kalman filter of
the project. I am trying to do a little program in Visual Basic that
reproduces a Kalman filter, simulating giroscope readings and injecting
random noise, so I can graphically analize the behaviour of the filter when
I interact changing in real time noise parameters (amplitude of spikes,
percentage of noise, positive noise vs negative noise) or changing the
values of the process noise covariance matrix or the measurement noise
covariance matrix. In this moment I'm fighting with the maths of part of the
process and I have a lot of doubts. I don't want to be pain asking a lot of
things, but I feel really stuck on some points.
Could someone give me a little orientation about this doubts:

-In revision 1.16 of ahrs.c is defined the constant dt, and it has a value
of 0.060. This constant, dt, is used at the time update (or propagation
state) of the Kalman filter. Is dt a factor that depends on the time, or on
how many times the filter is executed in a given time unit? Why is the ideal
0.032768 instead of 0.060?

-When the quaternion omega matrix is calculated, the notation that appears
on the old FAQ corresponds to Leibniz notation of derivatives? I mean,
d(q0)/d(q1) corresponds to the derivative of q0 in respect of q3 (in
spanish: la derivada de q0 respecto a q1)?

-And also: is there any estimation about the time when the new FAQ will be
avalaible, specially point 3 (Kalman filtering) ;-)

Thanks in advance for the answers and for this fascinating project.

Rafael H. Stark - ele...@ya...
"Creer es fácil, lo difícil es saber"



_______________________________________________________________
Copa del Mundo de la FIFA 2002
El único lugar de Internet con vídeos de los 64 partidos. 
¡Apúntante ya! en http://fifaworldcup.yahoo.com/fc/es/