The standard/usual way of estimating covariance matrices involves calculating sums involving the cross products of all pairs of variables. This is obviously greatly impacted by outliers. A single outlier on variable contaminates all the covariances involving . In order to estimate robust covariance matrices, we will use the robust package.