Standard estimation of covariance matrices is quite simple, as it relies on simple covariance and variance estimates for each variable. The usual formula for the covariance matrix does not discriminate whether the values are outliers or not (it involves taking the sums of the cross products between the observed vectors):
Robust estimation using covRob employs either the Donoho-Stahel or the orthogonalized quadrant correlation depending on the problem size. More algorithmic details can be found in the package's documentation: http://ftp.auckland.ac.nz/software/CRAN/doc/packages/robust.pdf.