What is the purpose of propagating the covariance in the prediction step?

• A. To eliminate all uncertainty before the measurement update.
• B. To propagate the current estimate uncertainty to the next step ahead of the measurement update.
• C. To adjust only the gravity model parameter.
• D. To reset the EKF state to zero.

Answer: (B)

The purpose of propagating the covariance in the prediction step is to forecast how uncertain the state will be at the next time step before any new measurement is incorporated. By applying the motion model (using its Jacobian to linearize the dynamics) and adding the process noise, you map the current uncertainty forward and obtain the predicted covariance. This predicted covariance expresses how confident you are about the forecasted state and sets up the Kalman update to balance the prediction with the incoming measurement. It’s not about eliminating all uncertainty, nor about adjusting a gravity parameter, nor resetting the state to zero.