An improvement over classical dynamic feedback linearization for the control of a nonholonomic mobile robot is proposed. The use of a state extension of higher dimension than in the case of dynamic feedback linearization helps reject constant disturbances on the rotational axis of the robot. The proposed dynamic extension acts as a velocity scheduler for the robot. It specifies at each time instant the ideal translational velocity that the robot should have. By having a two-dimensional state extension, both the magnitude and the orientation of the velocity vector can be generated, which accounts for improved robustness.