Intelligent Robotics Laboratory, Department of Systems Innovation, Graduate School of Engineering Science, Osaka University, Osaka, Japan.
PLoS One. 2022 Aug 10;17(8):e0271789. doi: 10.1371/journal.pone.0271789. eCollection 2022.
The emotion expressions of social robots are some of the most important developments in recent studies on human-robot interactions (HRIs). Several research studies have been conducted to assess effective factors to improve the quality of emotion expression of the robots. In this study, we examined the effects of a robot's vertical oscillation and transition on the quality of its emotion expression, where the former indicates the periodic up/down movement of the body of the robot, while the latter indicates a one-time up or down movement. Short-term and long-term emotion expressions of the robot were studied independently for the four basic emotions described in the circumplex model of emotions: joy, anger, sadness, and relief. We designed an experiment with an adequate statistical power and minimum sample size of human subjects based on a priori power analysis. Human subjects were asked to evaluate the robot's emotion expressions by watching its video with/without vertical movement. The results of the experiment showed that for the long-term emotions, the speed of vertical oscillation corresponded to the degree of arousal of the emotion expression as noted in the circumplex model; this indicated that fast oscillations improved the emotion expression with a higher degree of arousal, such as joy and anger, while slow or no oscillations were more suited to emotions with a lower degree of arousal, such as sadness and relief. For the short-term emotions, the direction of the vertical transition corresponded to the degree of valence for most of the expressed emotions, while the speed of vertical oscillation reflected the degree of arousal. The findings of this study can be adopted in the development of conversational robots to enhance their emotion expression.
社交机器人的情感表达是近年来人机交互(HRI)研究中的重要进展之一。已经进行了多项研究来评估有效因素,以提高机器人情感表达的质量。在这项研究中,我们研究了机器人的垂直摆动和过渡对其情感表达质量的影响,前者表示机器人身体的周期性上下运动,后者表示一次性向上或向下运动。我们分别研究了机器人的短期和长期情感表达,这四个基本情感是情绪的双因素模型所描述的:喜悦、愤怒、悲伤和宽慰。我们根据先验功效分析设计了一个具有足够统计功效和最小样本量的人类被试实验。人类被试通过观看机器人带有/不带有垂直运动的视频来评估机器人的情感表达。实验结果表明,对于长期情感,垂直摆动的速度与情绪表达的唤醒程度相对应;这表明快速摆动可以改善具有较高唤醒程度的情感表达,如喜悦和愤怒,而缓慢或无摆动则更适合具有较低唤醒程度的情感,如悲伤和宽慰。对于短期情感,垂直过渡的方向与大多数表达情感的效价相对应,而垂直摆动的速度则反映了唤醒程度。这项研究的发现可以应用于会话机器人的开发中,以增强其情感表达。