A Kalman Filter is a very effective type of filter that, in the case of VBOX’s, compares the positional and the velocity data as part of the smoothing routine. I.e. if the velocity exhibits a jump in speed over a short time yet the positional data does not corroborate this jump in speed, then the jump will be reduced and smoothed accordingly. This applies in the same manner for a jump in position that is not backed up by a jump in speed.
Live Kalman Filtering
Kalman Filtering that is applied live by a VBOX (where available), like any live filtering process, will not only smooth the data but will delay transients and sudden velocity changes in the data. This should therefore be used with caution, and only in cases where a live smoothed velocity or position is required from the VBOX.
Post processing Kalman Filtering
A Kalman filter that is applied to logged data has the benefit of being able to smooth individual data samples with respect to the values before and after the smoothed point. This has the major benefit of having virtually no effect on the latency of transients and velocity changes. This type of smoothing is available under ‘Tools’ in the VBOX Tools software.
Where possible, it is advised to log the raw data with as little smoothing as possible, and then apply post processing smoothing as required. If smoothing is applied at source, detail can not then be recovered.
The Position Quality channel, created by a VBOX is a numerical confidence indicator of the quality of the GNSS position based on the kalman filter estimated standard deviation. This channel is only present when IMU integration is enabled and will be logged, sent over serial or output via CAN on byte 7 of ID 308.
It is used for RTK robot path following applications, typically to verify position quality before starting a test.
- 10 = Difference between GNSS and KF estimated position is better than 5 cm.
- 9 = Difference between GNSS and KF estimated position is between 5–10 cm.
- 8 = Difference between GNSS and KF estimated position is between 10 cm and 1 m.
- 7 = Path following position confidence is between 1–2 m.
- 2 = When the IMU filter has been enabled and IMU has synced but is not yet initialised. Once initialised it should report as per the above values.
- 1 = When the IMU has not synced or if the KF has not been enabled (note if KF has not been enabled, then the value will only be present on CAN).
You can read more about Kalman filters in the following articles:
Quick Kalman Filter Initialisation
Benefits of IMU and Kalman filter integration for slip angle measurement