Human Visual Attention Model Based on Analysis of Magic for Smooth Human–Robot Interaction