I'm doing a kalman filter and I need to sum two covariance matrices (Q1 and Q2) regarding two uncertainties I need to join. I'm tracking the position of an object from a referencial with a changing position and orientation. As the position of the moving referential also has an uncertainty, it also has a covariance matrix. Both covariances matrices are 2x2 matrices, regarding the covariance in x, y, xy. So, the first covarience (Q1) refers to the uncertainty of the position of the moving object I'm trying to track. The second covariance (Q2) is the uncertainty of the position of the moving referencial in with I do the observation of the moving object. The total covariance will just be Q1+Q2, or do I need to do anything else?

  • 1
    $\begingroup$ I am confused on what you are asking. You can add two matricies of the same size because matrix addiition is well defined. But why are you doing that in the first place? $\endgroup$
    – Josh
    Apr 10, 2017 at 1:37
  • $\begingroup$ The nature of your question is becoming clearer, but it's still difficult to decipher. I suspect you will need to explain how the variables for the one covariance matrix are related to the variables for the other--otherwise, it will be impossible to know what needs to be done. $\endgroup$
    – whuber
    Apr 10, 2017 at 13:57
  • $\begingroup$ Sorry, again. I've added some more information. Thanks for your patience $\endgroup$
    – PR16
    Apr 10, 2017 at 20:43
  • $\begingroup$ Are the two sets of positions uncorrelated? That is, are the coordinates of the position uncorrelated with the coordinates of the moving "referencial"? Indeed, what is this--is it a set of coordinates parameterized by time or could it be an entire reference frame (coordinates and basic directions) that changes over time? $\endgroup$
    – whuber
    Apr 10, 2017 at 21:25
  • $\begingroup$ They are uncorrelated. And the coordinates ans positions can change over time $\endgroup$
    – PR16
    Apr 10, 2017 at 22:33


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.