# Revision history [back]

In Kalman filters, there are two stages to fusing every measurement: prediction and correction.

During the prediction stage, we predict from the current time to the time of the next measurement. The prediction uses the filter's kinematic model. Then, we correct the predicted state with the measurement data.

What you see is reflected in the code yu posted. The prediction stage isn't using the measurement itself, but the measurement's _time_ (measurement.time_). Then we correct the filter with the actual measurement itself.

In Kalman filters, there are two stages to fusing every measurement: prediction and correction.

During the prediction stage, we predict from the current time to the time of the next measurement. The prediction uses the filter's kinematic model. Then, we correct the predicted state with the measurement data.

What you see is reflected in the code yu you posted. The prediction stage isn't using the measurement itself, but the measurement's _time_ (measurement.time_). Then we correct the filter with the actual measurement itself.