Reduction of nonlinear control systems. A differential geometric approach. Upd. and rev. transl. of the Russian work. Transl. by P. S. V. Naidu (Q4240118)

A nonlinear control system in the author's sense is a system \[ \dot y= f_0(y)+ \sum^r_{j= 1}f_j(y) u_j,\quad y\in M\subseteq \mathbb{R}^n,\quad u= (u_1,\dots, u_r)\in \mathbb{R}^r,\tag{\(*\)} \] where \(M\) is interpreted as a space of states and \(u\) is a control. The vector fields \(f_j\), \(= 0,1,\dots, r\) are assumed to be smooth. The basic approach the author advocates, is to view the systems of type \((*)\) as the objects of a category such that morphisms between two such systems are maps between the state spaces which preserve solution trajectories. With such an interpretation one finds a natural notion of restrictions of control systems to (certain) subspaces of states. More importantly, the author also considers quotient objects (epimorphisms), which he calls reductions. The goal here is to partially decouple high-dimensional systems in order to solve the original problem by successively solving the quotient systems and plugging the solutions into the next quotient (viz. the Gauss algorithm or a composition series). Constructing reductions of this type and describing their properties is the main technical problem dealt with in this book. It is closely related to the construction of distributions (in the differential geometric sense) which are invariant under the vector fields \(f_j\). This explains why the author has to make heavy use of differential geometric concepts. He chose to include an introductory chapter of about 100 pages in which he explains all the prerequisites needed from (elementary) differential geometry. In spite of the categorical approach to control systems, the differential geometry is mostly explained in terms of local coordinates. The second half of the book is devoted to the introduction and description of the category of control systems, the notion of equivalence, and the basic concepts of restriction and factorization. In all cases, the general theorems (e.g. existence of certain normal forms etc.) are complemented by a large number of explicit examples.