Abstract:The robust Kalman filtering can resist errors from observation data or inaccurate state model that depend on the reasonability of equivalent weight. Usually the existed equivalent weight functions are defined as a segment function, which involves the selection of critical values for segment weight-decreasing. The selection is generally based on the confidence level of error tests or by experience, and it will never be changed once fixed. In order to further improve the efficiency of robust estimation, a new robust Kalman filtering with a continuous variable equivalent weight function is designed. Comparison and analysis has also been made between the new scheme and the common scheme. It can be concluded that the new designed scheme of robust weight determination has a better robust effect.