چكيده لاتين :
Least squares adjustment suffers by the outliers. In order to employ the least squares methods, it is necessary to be ensured that all of the observations are correctly measured. Before employing the least squares methods the pre-adjustment data screening should be performed, until the gross errors among the observations are detected. After doing least squares adjustment the post-adjustment data screening have to be employed to detect the small outliers in observations according to Baarda's theory. This process is valid for just one erroneous observation, it means that if there are several erroneous observations among the set of observations, one cannot correctly detect the outliers. In such cases the robust methods of estimation must be employed because it is insensitive to the outliers and gross errors. In this article among the various kinds of the robust estimation methods the method of Maximum Likelihood estimation of Huber will be considered in order to obtain the robust estimation of coordinates of points in a simple geodetic network, and the behavior of the coordinates of the net points due to the robust estimations in existence of up to six gross errors in observation are presented and convergence of the reweighting least squares is considered. It can be seen that to detect the outliers it is sufficient to investigate the residuals vector of the robust estimation. Also it can be shown that the root mean squares error between the robust estimation of coordinates of any points contaminated by six erroneous observations and the least squares excluding erroneous observations is about 0.3045 m.
