Higher-order techniques for some problems of nonlinear control (Q1886206)

Summary: A natural first step when dealing with a nonlinear problem is an application of some version of the \textit{linearization principle}. This includes the well-known linearization principles for controllability, observability and stability and also first-order optimality conditions such as Lagrange multipliers rule or Pontryagin's maximum principle. In many interesting and important problems of nonlinear control the linearization principle fails to provide a solution. In the present paper we provide some examples of how higher-order methods of differential geometric control theory can be used for the study of nonlinear control systems in such cases. The presentation includes: nonlinear systems with impulsive and distribution-like inputs; second-order optimality conditions for bang-bang extremals of optimal control problems; methods of higher-order averaging for studying stability and stabilization of time-variant control systems.