In this paper a method for the automatic command of three degrees of freedom of a robot using mouth gestures is presented. The method uses a normalized version of the a component from the CIELAB color space in order to detect the mouth region and to segment the mouth structures. A set of features extracted from the segmented regions serve to model a set of activation signals, who control the transitions between states in a state machine that translates gestures into actions. Results and conclusions of the behavior of the system are presented.