On the largest feedback linearizable subsystem (Q1066092)

The problem of partial linearization of an analytic affine nonlinear control system \(\dot x=f(x)+g(x)u\) is considered. In this problem one tries to identify (i) a state feedback \(u=\alpha(x)+\beta(x)v\) and (ii) a state space transformation such that the feedback modified system in the new state space coordinates \((\tilde x_ 1,\tilde x_ 2)\) is partially linear; that is the \(\tilde x_ 1\)-dynamics is simply a linear control system. Obviously this opens the way of partially applying linear control techniques. For practical purposes it is of course desirable to make the linear part (the \(\tilde x_ 1\) coordinates) as large as possible. In the present paper the author gives a straightforward explicit way of computing the largest linearizable subsystem. This subsystem and its dimension can be computed by means of standard geometric techniques in nonlinear control. When the largest linearizable subsystem lives on the whole state space the results reduce to that of \textit{B. Jakubczyk} and \textit{W. Respondek} [Bull. Acad. Pol. Sci., Ser. Sci. Math. 28, 517-522 (1980; Zbl 0489.93023)]. We note that the same problem is studied by \textit{W. Respondek} in ''Partial linearization, decompositions and fibre linear systems'', Proc. MTNS Conf., Stockholm 1986 (to appear).