The first impression of robot appearance normally affects the interaction with physical robots. Hence, it is critically important to evaluate the humanoid robot appearance design. This study towards evaluating humanoid robot design based on global eye-tracking metrics. Two methods are selected to extract global eye-tracking metrics, including bin-analysis-based entropy and approximate entropy. The data are collected from an eye-tracking experiment, where 20 participants evaluate 12 humanoid robot appearance designs with their eye movements recorded. The humanoid robots are evaluated from five aspects, namely smartness, friendliness, pleasure, arousal, and dominance. The results show that the entropy of fixation duration and velocity, approximate entropy of saccades amplitude are positively associated with the subjective feelings induced by robot appearance. These findings can aid in better understanding the first impression of human-robot interaction and enable the eye-tracking-based evaluation of humanoid robot design. By combining the theory of design and bio-signals analysis, the study contributes to the field of Transdisciplinary Engineering.
Transfer Learning for Humanoid Robot Appearance-based Localization in a Visual Map
