“…The method of combining multiple filters according to an innovation-based criterion is the key of our estimation method. The Kalman filtering algorithm, which is one of the most popular algorithms [23,63,100,112], is to effectively reduce Gaussian white noises, while the H ∞ filtering one [33,85], which has been becoming popular, e.g., [6,27,47,49,59,78,106], can accommodate norm-bounded deterministic noises, and therefore is expected to be robust against model errors. Our method of combining filters is to utilize the advantage of each filter as much as possible.…”