Ellipsoidal calculus for estimation and control (Q1349390)

This is an original and well written book devoted to problems of estimation and control of uncertain dynamical systems governed by ordinary differential equations. The information about uncertainties is of nonstochastic nature: only bounds on the magnitude of some variables are known. Such uncertainties may appear in the initial state, in the observation or as an unknown input in the dynamics -- related to differential games. Therefore the relevant mathematical treatment is based on set-valued analysis. The book consists of four parts. The first part presents the theoretical base of the ``guaranteed'' approach to such problems as reachability, solvability, viability and control synthesis, possibly in presence of control and state constraints and uncertainties. The main tools are the so-called funnel equations describing the dynamics of the relevant sets in the state-space on which a feedback design is based. The links with Hamilton-Jacobi-Bellman equations are given through the level sets of their solutions, giving another view on the dynamic programming approach. The book deals with linear systems but some of the results in this part could be extended to nonlinear systems. The extension to the nonlinear case of the numerical treatment, however should be more complicated. In order to handle set-valued solutions numerically, the authors choose to approximate sets by ellipsoids. In the second part of the book, they develop a thorough ellipsoidal calculus which is certainly interesting for other areas, in particular in scientific computations. Sums, (Minkowski) differences and intersections of ellipsoids are represented or approximated by parametrized families of ellipsoids. Optimal ellipsoidal approximations are obtained with respect to a number of meaningful criteria. The third part of the book interprets in a constructive way the theory, developed in the first part, by using ellipsoidal-valued dynamics as approximations of the funnel equations. This gives rise to inner and outer approximation of attainability and solvability tubes under state constraints and disturbances, and respective feedback design. Many examples are worked out with nice computer graphics. More advanced issues as links with \(H^{\infty}\)-approaches, tracking problems under uncertainty, singular perturbation techniques, etc., are considered in the fourth part of the book. The book is of particular interest for researchers, applied mathematicians and engineers in the field of control theory, but could be also used by a broader audience of readers for acquaintance with a constructive technique for the ``guaranteed'' approach to uncertainties -- a modern complement to the stochastic one.