This paper describes a robotic-head system as a multimodal communication device for human-robot interaction, and the system's potential application in home environments. Most robotic systems for natural user interaction have facial expressions, since facial expressiveness is regarded as a key component to developing personal attachment along with prosodic expressiveness. In the first part of the paper is the description of our robotic head system Character Robot Face (CRF). A deformation approach and a parametric normalization scheme are proposed to produce facial expressions of nonhuman face models with high recognition rates. In the second half of the paper, CRF is endowed with artificial emotions and assigned tasks conceivable in home environments. A coordination mechanism between the robot's mood (an activated emotion) and its task is proposed so that the robot can, by referring to the emotion-task history, select a task depending on its current mood if there is no explicit task command from the user. When the robot performs a task, a particular emotion value gets boosted according to the same emotion-task history so that the emotion is more likely to be activated.

This paper explores a human-robot mutual communication system, which human users can either communicate with or use as an information terminal. In particular, we propose the deformation based facial expression system. We also propose a robotic vision system, which changes its visual attention according to the environment. Firstly, the system must have advanced abilities to express their intention by means of making facial expressions, gestures, or speech. Above all, facial expression. Reconsidering the facial action coding system and action unit from the point of what expression of the robot human recognize easily. We propose the deformation based expression system. Secondly, to realize fluent communication between human and robots, we propose a robot vision system changing its gazing communication according to the environment and situation based on visual recognition. We developed an original character robot (CR) and evaluated the proposed methods. Consequently, it was shown that human-robot mutual communication is achievable.