A linear Kalman filter has a simple geometric interpretation in terms of vector projections. [br]This applet explorers the geometry behind a recursive estimation of a random variable x from two noisy measurements [math]z_1[/math] and [math]z_2[/math] . [br]Geometrically, we obtain the orthogonal projection of x onto [math]z_1[/math] followed by the orthogonal projection of the result on the plane spanned by [math]z_1[/math] and [math]z_2[/math].[br]You can grab the picture with a mouse and move it around to have a better look.[br]See the attached text for details.