In a Kalman Filter, what is the specific role of the 'prediction' step when tracking an object through temporary occlusion?

In a Kalman Filter, the prediction step serves to maintain an estimate of an object's state when direct observations are missing due to occlusion. The Kalman Filter operates in a continuous loop consisting of a prediction phase and a correction phase. During the prediction phase, the filter uses a mathematical model of motion, such as constant velocity, to calculate the object's exp....