Enhanced Facial Expressions of Avatars by Internal Movable Weights

When recognizing the facial expressions of avatars or virtual agents, visual information is used. We developed a device that can move internal weights according to facial expressions of a virtual agent on a screen. The device allowed the user to perceive the facial expressions with visual and tactile information. We conducted a pilot test, and the results suggested that moving the internal weights in accordance with the facial expressions enhanced the credibility of the virtual agent. We also created a future scenario where the device will be used.