On robust estimation with correlated observations

Robust estimation methods have been widely used in Photogrammetry and Geodesy. They have an implicit basic premise that all observations are uncorrelated. However, we have to deal with various types of correlated observations in surveying practice. This paper therefore extends robust estimation methods to the case of correlated observations with the help of bivariate functions. The properties of the solution have been investigated and the proper confidence region problems have been discussed. A simulated example shows the applications of the proposed methods.