Abstract
Recently, the research on non-verbal information has been actively performed in the human symbiotic system research. The authors also perform the study on viewpoint recognition in this area. In this paper, Starburst algorithm has been used. This algorithm is able to track the pupil by recognizing corneal limbus contour (the contour between the iris and the sclera), which is the most important feature of the eye image in the visible spectrum. We also present a low-cost and Kansei input device with both viewpoint measurement and HMD environment display function by utilizing the front camera and display of a smart tablet. Moreover, we have designed a teaching system based on the vision image from a mobile robot by using this system which behavior of the mobile robot can be taught by the instruction via human's eye remotely.
