Adaptive Nonlinear Control With Contraction Metrics

Abstract

This letter derives direct adaptive control algorithms for nonlinear systems nominally contracting in closed-loop, but subject to structured parametric uncertainty. The approach is more general than methods based on feedback linearization or backstepping as it does not require invertibility or the system be in strict-feedback form. More broadly, it can be combined with learned controllers that must remain effective in the presence of structured parametric uncertainty. Simulation results illustrate the approach on a system with extended matched uncertainty.