Regularized Robust Estimation withApplications to Wearable Computing

There are always discrepancies between design models and the actual physical systems or phenomena that they model. Regardless of their source, such perturbations can degrade the performance of otherwise optimal designs. This talk describes a design strategy for state-space models with bounded perturbations. In comparison to other robust formulations, the resulting procedure performs data regularization as opposed to de-regularization; a property that avoids continuous testing of existence conditions and is therefore useful for applications involving online/real-time operation. An application in the context of mobile wearable computing is described. Wearable computers are devices that are supposed to provide users with real-time access to information databases in a natural and unobtrusive manner. A critical factor for the success of wearable computers is their ability to quickly and reliably process sensory data, as well as their ability to interact reliably with the user. This talk uses the developed robust filtering theory to describe a robust hand-gesture interface for wearable computing applications. Emphasis is placed on the issues of adaptivity and robustness, since such devices are expected to operate reliably under changing conditions.