(see KalMOOC)
The robots can measure angles (with a goniometer) from landmarks or from the other robot if the distance is small. Using these measurements, each robot can estimate its position and we show how their covariance evolve.
A robot is localized and tracked from two noisy radars using an EKF.
An inverse pendulum is regularized using an EKF for state observation.
A Kalman filter is used to predict the robot position and correct it by detecting seamarks. At the end a backward pass, is used to further refine the seamarks and the robot positions.