Choose error covariance Q and R in Extended Kalman Filter

I'm designing an Extended Kalman Filter for a nonlinear system. The objective is to estimate an unkown parameter (not a state variable). The problem is with each Q and R, I receive another responsed estimation. I think Q and R are similar to coeffiecents in PID controller. Do you have any documents or suggestions for tuning these parameters?

Answer 1

[pham Khanh SOn]

Hi, Q R is not the similar to coefficents in PID controller. You can check out Matlab's tutorial on Kalman Filter.