Robust state and unknown input estimation: novel algorithms for practical use and predictive control schemes

This thesis deals with the development of robust state estimators for mitigating the impact that unknown inputs have on state estimates. For nonlinear systems, robust state estimators are developed, considering tradeoffs involved in deployment/implementation for real-world applications. Recursive strategies are designed to estimate the unknown inputs along with the states. Considering practical case studies (involving the rail track-geometry inspection vehicle, localization of a robot, and the quadruple tank system), we show how the trinity of the development of a state-space realization, the robust state estimator, and the unknown input estimator helps enable robust sensor fusion exercise.