Approximate models for adaptive feedback linearization (Q4374754)

The authors generalize a result of \textit{S. S. Sastry} and \textit{A. Isidori} [IEEE Trans. Autom. Control 34, No. 11, 1123-1131 (1989; Zbl 0693.93046)], on adaptive stabilization and tracking of affine systems with a well defined global relative degree and linear parametric uncertainty. Through a modification of the feedback law, robustness under output disturbances is achieved. This allows to deal with the error introduced when approximating the original system. The focus is placed on linear approximation schemes with compactly supported basis functions. It is remarkable that in contrast to many other neurocontrol results the control is a pure state feedback law, in particular no assumptions on the availability of the output Lie derivatives are made. Even though the most general results can only be local in nature due to the choice of the approximation scheme, global versions are obtained under some additional assumptions already used by Sastry and Isidori.