I am measuring position and velocity, both have some noise in them. Velocity is defined as derivative of position. The system is apparently non-linear so I need to use EKF.


\begin{align} X&=\begin{bmatrix}p\\v\end{bmatrix}\\ p_{k+1}&=A_0\sin\left(b_0k\right)+A_1\sin\left(b_1k\right)+A_2\sin\left(b_2k\right)\\ v_{k+1}&=A_0b_0\cos\left(b_0k\right)+A_1b_1\cos\left(b_1k\right)+A_2b_2\cos\left(b_2k\right)\\ \end{align}


  1. I suppose this is a non-linear system. Am I right?
  2. If yes, I should use Extended Kalman filter. But how do I get $f$ and $h$ functions for my case (below)? The current state doesn't depend on previous state; except the $k$ (step) variable. \begin{align} x_k&=f\left(x_{k-1}\right)+w_k\\ z_k&=h\left(x_k\right)+v_k \end{align}
  • $\begingroup$ Your current state should depend upon its previous state, or a Kalman filter isn't going to help you much. Try solving for $p_{k+1}$ in terms of $p_k$ and so on. This would yield the state transition equation, which is an important part of the model. $\endgroup$
    – Jason R
    Aug 13, 2012 at 17:19
  • $\begingroup$ Thanks for your input. I was concerned this wouldn't be applicable for Kalman's filter. $\endgroup$
    – c0dehunter
    Aug 13, 2012 at 19:12


Your Answer

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

Browse other questions tagged or ask your own question.