Imagine driving a car with a GPS and a speedometer. You know your starting position, but as you drive, the GPS occasionally updates your location. Meanwhile, you can guess where you will be based on your steering and speed, but both guesses and measurements have errors. How do you blend them in the best possible way to know where you are right now, and say how sure you are? That is exactly what the Kalman filter does for any moving system — it gives you the best guess of the hidden states and their uncertainty, updating every time new data arrives.