Inertial Navigation Integration FIlter

[catalin.drago]

Hello,
I am trying to create a filter that can integrate inertial measurements with GPS measurements. My inertial sensors are a yaw axis gyro, an odometer, and possibly a 3axis accelerometer.
Could you please help me with some guidance? I have read papers, articles and books on the subject but I am still a little confused.
I consider the acceleration constant, so I have a state vector x=[E N v_E v_N a_E a_N], E and N are the positions in NED coordinate system, v_E and v_N are the speed on each axis, and a is the acceleration on each axis
The measurement vector z is [E_GPS N_GPS E_DR N_DR a_E a_N] where E_GPS and N_GPS are the coordinates obtained from the GPS receiver and E_DR and N_DR are the coordinates obtained from the inertial sensors through dead reckoning, and a is the acceleration obtained from the accelerometer.
F, the transformation matrix is: [1 0 dt 0 (dt^2)/2 0; 0 1 0 dt 0 (dt^2)/2; 0 0 1 0 dt 0; 0 0 0 1 0 dt; 0 0 0 0 0 1];
H, the measurement matrix is: [1 0 0 0 0 0; 0 1 0 0 0 0 ; 1 0 0 0 0 0; 0 1 0 0 0 0; 0 0 0 0 1 0; 0 0 0 0 0 1],
Does it make any sense so far? Should I chose a different state vector?
My problem is that I don't know, for this system how can I determine the process noise covariance and measurement noise covariance.
Can you give me some indications in that direction?
Thank you very much.

 

[mugaliens]

http://en.wikipedia.org/wiki/Kalman_filter" .

 

[catalin.drago]

I passed the level of reading wikipedia.
I was hoping for help, not links to wikipedia

 

[catalin.drago]

I was hoping for some pointers from someone who has experience with the implementation of Kalman filters. I would like to know if that is a well designed filter, or if I should introduce more elements in the state vector like heading, and odometer and gyro bias, or the filter is going to work well also without them?