Systems experiencing events in the order of 10μs-10ms timescales, for instance highrate dynamics or harsh extreme environments, may encounter rapid damaging effects. If the structural health of such systems could be accurately estimated in a timely manner, preventative measures could be employed to minimize adverse effects. Previously, a Variable Input Observer (VIO) coupled with a neuro-observer was proposed by the authors as a potential solution in monitoring their structural health. The objective of the VIO is to provide state estimation based on an optimal input space allowed to vary as a function of time. The VIO incorporates the use of mutual information and false nearest neighbors techniques to automatically compute the time delay and embedding dimension at set time intervals. The time delay and embedding dimensions are then used to vary the input space to achieve optimal performance for the estimator based on the observed measurements from sensors. Here, we augment the VIO with a smooth transitioning technique to provide enhanced robustness. The performance of the algorithm is investigated using experimental data obtained from a complex engineering system experiencing a harsh extreme environment. Results show that the enhanced VIO incorporating a smooth transitioning input space outperforms the previous VIO strategies which allowed rapid input space adaptation.