Feedback linearization via state transformation using estimated states