New results in robust location estimation with trimmed averages

Abstract

When there are more than necessary distance measurements in localization by distance measurements with closed form estimators, forming smaller subgroups of measurements and averaging the location estimates obtained with these subgroups of measurements makes it possible to eliminate outlier measurements if they are present. In order to eliminate these outlier results, the nearest estimate to the geometric median of estimates is proposed as a reference in this work. Conducted simulation studies show that significant gains can be obtained using geometric median in place of arithmetic average in robust averaging methods.