I'm getting this warning from Matlab about Kalman Gain.

Warning: Matrix is close to singular or badly scaled.
         Results may be inaccurate. RCOND = 9.996841e-19. 

The problem is coming from high variance of the measurement model. My question is here Does EKF work with high noise in sensor?

  • 1
    $\begingroup$ It is hard to tell what went wrong here without more information. My guess is that inverting the innovation covariance matrix ($H \bar\Sigma_t H^T + Q_t$) fails due to $Q_t$ being degenerate. EKF does work well with high noise, as long as this noise is still Gaussian. Could you provide the values of your matrix $Q$ and maybe even $H \bar\Sigma_t H^T$ in this case? $\endgroup$
    – sebsch
    Commented Jun 23, 2014 at 14:18


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.