I used the measured position to estimate the velocity using a Kalman filter.

But in simulations, the Kalman gain changes quickly and then remains constant when position and velocity continue to change. For example, position and velocity change at times 0 to 0.5(s) and 3 to 4(s), respectively. But the Kalman gain only changes 0 to 0.1(s) and then remain constant.

My question is: Why is the Kalman gain constant when position and velocity change?


The Kalman Gain $K_k$ for a discrete, linear system is computed using the state covariance matrix $P_k$, measurement matrix $H_k$ and measurement noise covariance matrix $R_k$:

$$ K_k = P_kH_k^T(H_kP_kH_k^T+R_k)^{-1} $$

If the measurement matrix and the measurement noise covariance matrix are constant and the state covariance matrix converges to a steady-state value $P_\infty$, the Kalman Gain also converges to a steady-state value $K_\infty$, regardless of actual measurements or state.

| improve this answer | |

Your Answer

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

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