4-dimensional Kalman filter demo

Click anywhere on the page to add measurements. The green circle is your predicted next location.

Keep in mind that the timestep between clicks is fixed, meaning it doesn't matter how often do you click. Your next click always provides the location at t = t + 1

The Kalman filter expects a linear movement with a constant velocity. So one thing in particular that you can see is how poorly the filter reacts to sudden changes in movement direction.

Made by Leonid Shevtsov for the "CS 373: Programming a Robotic Car" course at Udacity.