Ellipsoidal state estimation based on sum of squares for non-linear systems with unknown but bounded noise

This brief concerns the set-theoretic recursive state estimation of non-linear systems with polynomial dynamics, making use of ellipsoidal bounds. The sum of squares approach is used to construct an optimal ellipsoid containing the intersection of sets obtained with the use of non-linear system equations and measurement equations. The properties of the proposed filter are illustrated with two examples, one concerning the simultaneous estimation of parameters and state of a linear system, and the other on the Euler equations of the rigid body rotational motion.