Nonlinear robust and adaptive control with application to brake control for automated highway systems

The work presented in this dissertation was motivated by the need of accurate vehicle longitudinal control in an Automated Highway System (AHS) environment. Due the model and parameter uncertainties present in the brake system as well as many other physical systems, robust and adaptive control algorithms are investigated in this dissertation.; In order to provide a physical motivation for the later theoretical studies, a powertrain and brake system model are developed. These models provide the basis for both the model-based control algorithms developed here and the simulations resulting from their investigation. Due to the nonlinearities and uncertainties inherent in such systems, Sliding Mode Control is employed to control the longitudinal behavior of the vehicle. Although good results are obtained using this method, improvements can be obtained by using robust and adaptive control algorithms.; The method Dynamic Surface Control (DSC) is extended to include nonsmooth functions. This method formalizes the concept of multiple surface sliding mode control and the use of synthetic inputs and filters. Filippov's theory of differential inclusions is used to prove the stability of these algorithms. The advantage of such a formulation is reduced final tracking error.; When employing a DSC algorithm, it is assumed that the entire state is available for measurement. When this is not the case, an observer is used to estimate the states. Combining the two yields a nonlinear sliding compensator whose asymptotic semi-global stability is shown.; When there is uncertainty in the value of a certain parameter, adaptation and estimation may be used to determine the correct value of the parameter. Therefore nonlinear adaptation and estimation methods are developed. The novelty resides in the use of nonsmooth Lyapunov functions which lead to nonsmooth parameter adaptation (estimation) laws. The advantage of such methods is parameter convergence in finite time, adaptation (estimation) even when the error is small and less stringent requirements on the frequency content of the input.; These methods, generic in nature to be applicable to many physical systems, are used to improve the control of automotive brake systems. It is thus shown that these algorithms are implementable even to systems as complex as an automobile. Simulation and experimental results are used to corroborate the theoretical findings of this work. They both show velocity tracking improvement at no additional input cost.