I'm trying to build a model based on tracking an object in 2-D using MATLAB (time-varying system). I built the model using an Extended Kalman Filter and a PID controller.

I have a doubt regards the relation of the reference with. I think that it is a theoretical doubt but, after some research I still do not understand.

In your opinion what are the signal that must follow my reference? The state ("real" position x and y) or the estimated state (estimated position x and y")?

Thank you in advance.

  • $\begingroup$ Do you know the true state (or at least the real position x and y)? $\endgroup$
    – fibonatic
    Commented Sep 21, 2022 at 21:17
  • $\begingroup$ I know the initial position regarding x and y. $\endgroup$
    – marcck
    Commented Sep 22, 2022 at 9:55
  • $\begingroup$ Welcome to Robotics marcck, but I'm afraid that it is not clear what you are asking. We prefer practical, answerable questions based on actual problems that you face, so it's a good idea to include details of what you want to achieve, what you tried, what you saw & what you expected to see. Please take a look at How to Ask & tour for more information on how stack exchange works and work through the Robotics question checklist to edit your question to make it clearer. $\endgroup$
    – Ben
    Commented Sep 23, 2022 at 14:05
  • $\begingroup$ I was editing my question, but a great answer arrived. However I'll take into account these guidelines. Thanks. $\endgroup$
    – marcck
    Commented Sep 28, 2022 at 18:24

1 Answer 1


You always act on the estimated state, because if you knew the actual state then you wouldn't need an estimator!

Since you're in simulation, you could run the same test several times, running the PID controller with the real state one time and with the estimated state the other time. The results should be basically identical, and it's a good way to prove to yourself (and others!) that your estimator and complete system (estimator + controller) are working well.

I'll add though that you should run whatever tests you can with the actual system (step test, disturbance injection, etc.) and re-run the same tests against your modeled system. A great estimator and great controller don't really matter if your underlying model doesn't actually reflect reality!

  • 1
    $\begingroup$ You're right, thanks a lot! $\endgroup$
    – marcck
    Commented Sep 28, 2022 at 18:25

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.