Making eyes with robots

摘要

One of the most important aims of social robotics is to improve Human-Robot Interaction by providing robots with means to understand observed behavior and to predict upcoming actions of their in-teraction partners. The most reliable source for inferring the action goals of interaction partners is their gaze direction. Hence, to anticipate upcoming actions, it is necessary to identify where others are currently look-ing at and to shift the attentional focus to the same location. Interestingly, it has been shown that observing robot gaze direction also induces attentional shifts to the location that is gazed-at by the robot. Given this, gaze direction can be actively used by the robot to direct the attentional focus of interaction partners to im-portant events in the world. In this paper, we review findings from two studies indicating that the readiness to engage attentional resources in interactions with robots is modulated by the degree to which intentionali-ty can be attributed to the robot: Robots believed to behave similar to humans cause stronger gaze-cueing effects than robots perceived as machines, independently of their physical appearance. Based on these find-ings, and on results from a pilot study with a sample of patients diagnosed with autism spectrum disorder (ASD), we derive guidelines for improving human-robot interaction by emulating social gaze behavior.