milsto / robust-kalman Goto Github PK

Is there a specific reason for not using the author's method for computing the new state variables? Like problems during matrix inversion etc.?

For example, sometimes the inverse of the matrix that is pointed out in the image is not computable due to singularity. Or am I doing something wrong? Could you help me with that, please? Thanks!

Also, I am implementing this based on the EKF model but I believe that should not be an issue?

I'm building a weighing system. Here I filter analog data with kalman filter and moving average. I have a problem like this.

I put 2 kg on the scale and shake the weighing mechanism and create vibration. In this case, the 2 kg value moves +- 20 grams. If I increase my filtering coefficients, this play disappears, but this time the system reacts very slowly to changes of 1 gram. How can I solve this problem can you help me.