Autofocus control in the visual system of an avatar robot

Abstract

This study develops a visual system for an avatar robot. The system uses eye-tracking data obtained from a head-mounted display to perform autofocus control on images captured by a stereo camera. The distance between the two cameras of the eyes of the avatar robot is automatically controlled according to the interpupillary distance of the operator measured by the head-mounted display (HMD). The distance to the point of fixation is determined through distance estimation using eye tracking of the stereo camera, and the focus is adjusted to that point. Distance estimation was performed using geometric calculations. In the experiment, subjects wore the HMD, and virtual reality (VR) sickness was evaluated using subjective assessment methods; simulator sickness questionnaire (SSQ), virtual reality sickness questionnaire (VRSQ), and fast motion sickness scale (FMS). The superiority of autofocus was evaluated using the Wilcoxon signed rank test. Most values were less than 0.05, indicating that autofocus control is effective in reducing visual fatigue and mitigating VR sickness. This advancement enhances depth perception, reduces VR sickness, and improves immersion during operation.