A Kalman filter is an advanced smoothing algorithm used in VBOX systems to improve the stability of position and velocity data.
The filter compares changes in position and velocity data to identify and reduce sudden spikes or inconsistencies. For example, if the velocity data shows a sudden change that is not supported by a corresponding change in position, the filter reduces and smooths the anomaly. The same process is applied to positional changes that are not supported by velocity data.
Live Kalman filtering is performed directly by supported VBOX units during data capture.
As with all live filtering methods, the process not only smooths the data but can also introduce a delay in sudden changes or transients within the recorded velocity and position data.
For this reason, live Kalman filtering should only be used when smoothed live position or velocity data is required.
IMPORTANT
Once smoothing has been applied during logging, the original unsmoothed detail cannot be recovered.
Post-processing Kalman filtering is applied after data has been logged.
Unlike live filtering, post-processing can analyse data samples using surrounding values recorded before and after each point. This significantly reduces the latency normally introduced by live smoothing methods, while still improving data stability.
TIP
Where possible, record raw data with minimal smoothing applied and perform additional smoothing during post-processing as required.