Strong Tracking Filters: Derivation and Improved Heuristic

摘要

This paper reviews variants of “strong tracking filters,” and offers a new method of adjusting the innovation matrix that allows the filter to be used with measurements of varying dimensionalities over time. Strong tracking filters are more robust to model mismatches than Kalman filters. Variants of the extended strong Kalman filter are compared to the extended Kalman filter for two types of model mismatch.