Abstarct: The estimation of regression parameters is very important in analysing and predicting data‎. ‎A common method is the least squares estimator that it does not have good performance in case of multicollinearity, ‎so that the shrinkage estimators were born‎. ‎One of these most popular techniques is bridge estimator‎. ‎Also‎, ‎the least squares and shrinkage estimators are influenced by outliers‎. ‎Although‎, ‎there are many papers about these estimators‎, ‎the suggested methods are not good when the data suffers multicollinearity problem‎. ‎In this dissertation‎, ‎we propose a new method called robust bridge that controls the existence outliers and high multicollinearity‎. ‎Also‎, ‎the asymptotic behavior of this estimator is studied and the computational algorithm is introduced‎. ‎As a special case‎, ‎the robust LASSO and ridge and their ‎fuzz‎ied versions are verified‎.