In this paper the authors report on the creation of a face tracking function that combines foveal vision and peripheral vision for communications between a humanoid robot and a person. In the authors' approach, by combining foveal vision and peripheral vision using a particle filter, a function for robust face tracking and attention direction can be achieved. In particular, the authors used peripheral vision when the robot was looking at an object other than a human being in order to make a deliberate gesture, and were able to achieve robust face tracking at all times by combining foveal vision and peripheral vision when the robot was paying attention to the person's face. Moreover, robust face tracking can also be applied to conventional technologies for facial expression recognition and utterance detection during interactive communication. © 2006 Wiley Periodicals, Inc.