Robust Kalman Filtering based on Multiple Hypothesis Techniques

In this paper, a new robust state estimator for linear systems with parametric uncertainties is proposed. The uncertainties affecting the system are regarded as unknown sequences of quantized parametric uncertainties. And then the exact robust estimator is derived by handling the uncertainty parameter sequence hypotheses by means of multiple hypotheses testing (MHT) techniques. However, since the exact filter has to treat ever expanding hypotheses, a suboptimal estimator based on zero scan back concept is proposed. A benchmark example for robust Kalman filtering is demonstrated to compare the performance of the proposed filter with those of an existing robust Kalman filter