A Control Strategy of Robot Eye-Head Coordinated Gaze Behavior Achieved for Minimized Neural Transmission Noise

Abstract

Many studies have demonstrated the necessity to drive robot displaying natural and appropriate behaviors in social scenes. In this article, efforts were taken to restore the mechanism of human gaze behavior that can be highly informative for robot-human interaction. In order to determine the rules that biological plants obey in gaze behavior, we modeled the eye-head coordinated gaze behavior as a two degree of freedom synthetic system, and obtained a closed-form equation for determining the movement duration and dynamics of it. By solving the equation of this model numerically under the condition of minimal neural transmission noise effect, it was found that this model can reproduce the gaze shift behavior and predict the coordinated trajectories of eye movement and head torsion. The proposed model and methodology was tested on the Xiaopang robot platform. By directly comparing the experimental result with the practical observation data, it indicates that the proposed model and methodology is robust to represent the pattern of human eye-head coordinated gaze behavior, this concludes that the human gaze sequence has evolved as a strategy to optimize the tradeoff between focal fixation accuracy and gaze shift speed.