Improved transient performance of nonlinear adaptive backstepping using estimator resetting based on multiple models

It is proved that the the transient performance of nonlinear adaptive backstepping can be improved by re-setting the parameter estimator, without loss of stability. The estimator re-setting algorithm is based on multiple-model adaptive control, where a number of models with fixed parameter vectors are monitored online in order to detect a parameter vector that gives a negative jump in the control Lyapunov function when replacing the estimate provided by the standard adaptation law. An application to wheel slip control is studied.