Re: [Paparazzi-devel] Problems with ahrs Kalman

[Prof. Dr.-Ing. Heinrich Warmers]

Hi,
i have no practice with the Paparazzi filter but i saw the following with our implementation of a EKF:
The results are affected of the data types (32 bit  real or 64 bit real).  We had differences with the mathlab model (64 Bit) and the  implementation  on the lpc2148 (32 Bit).
The micormag  has often very bad data and a lot of spikes.
 Are you sure that the implementation can work with all  angles?
In the video there were only  movements in the range of +-45 grad.
A question to the developer group was  never answered.
By our implementation of the complementary DCM  filter  (Spring 2010) we  had to improve the Paparazzi code to became
compatible with all angles ranges.
Sometime you get gimbal  locks.

If you want to have a  look to the problems please read:
http://code.google.com/p/gentlenav/downloads/list

Regards

Heinrich


Patrick McColl schrieb:

Hi,

 

I'm using the ahrs-Kalman code  (paparazzi3 version from koders.com) on a board of my own design.  The board uses Rabbit BL4100 microprocessor board and Analog Devices ADIS16364 gyro/accelerometer and Micromag3 magnetic sensors.

 

The code works fine with small movements but turns unstable after a minute (or so) when I move the board simulating a normal flight.

 

Has anybody experience the same problem?  Is there any way to fix it?

 

While debugging I noticed that the P matrix is symmetric all the time. Is this correct? 

If so, the code can be optimized by calculating only half the matrix and just filling the other half with the calculated values.

 

Thanks,

 

Patrick 

 

_______________________________________________
Paparazzi-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/paparazzi-devel