It would be useful to have an option to exclude measurement noise when computing statistics of the predicted measurement. In particular, we've been interested to use the predict_measurement method of UnscentedKalmanUpdater when performing measurement linearisation for the IPLF algorithm. However, the unscented transform inside predict_measurement always receives noise covariance, whereas it could have worked without it if covar_noise=None is used, as enabled by this feature in sigma2gauss.
Look like it shouldn't be difficult to add an additional Boolean argument (noise/measurement_noise/covar_noise??) to all the updaters' predict_measurement method to support this.
It would be useful to have an option to exclude measurement noise when computing statistics of the predicted measurement. In particular, we've been interested to use the
predict_measurement
method ofUnscentedKalmanUpdater
when performing measurement linearisation for the IPLF algorithm. However, the unscented transform insidepredict_measurement
always receives noise covariance, whereas it could have worked without it ifcovar_noise=None
is used, as enabled by this feature insigma2gauss
.