Hi Swift Team,
I'm fusing the GNSS readings into a Kalman Filter (KF) for robot localisation.
With reference to the following tutorial:
I am reading off h_accuracy and v_accuracy in LLH messages to generate the covariance matrix for the GNSS readings. I've read that the h_accuracy and v_accuracy values are derived dynamically from the internal covariance calculation which is based on the quality of the underlying measurements (pseudorange, acceleration, etc), and the degree to which the new measurements agree with the latest estimates.
Can I have more details on the formulas for h_accuracy and v_accuracy or some white-paper on how the values are derived (preferably for a moving rover GNSS antenna) as I'm tuning my KF (and I need to ensure the reasonableness of the h_accuracy and v_accuracy values)?
P.S: I'm using a Piksi Multi GNSS Module.
Thanks!
Best regards,
Md Hizam