Imitating Human Emotions with Artificial Facial Expressions
Imitating Human Emotions with Artificial Facial Expressions
复制标题
DOI：
10.1007/s12369-013-0211-1
复制
发表时间：
2013-11-01
影响因子：
4.7
通讯作者：
van der Pol, David
中科院分区：
文献类型：
--
作者：
Johnson, David O.;Cuijpers, Raymond H.;van der Pol, David
人类通过面部表情来传达有关他们情绪状态的信息。机器人通常不能像人类一样表现出面部表情，这使得它们很难模仿情感。在这里，我们研究如何在Aldebaran的Nao机器人眼睛周围的LED图案可以用来模仿人类的情感。我们进行了两个实验。在第一个实验中，我们研究了LED的颜色，强度，频率，锐度和方向，人类与不同的情绪。根据结果，创建了12个LED图案。第二个实验测量人类如何识别这些LED图案作为设计意图的情感。我们使用ROC（受试者操作特征）图来确定12种LED模式中哪一种是Nao机器人模仿情感的最佳模式。我们使用ROC图的技术通常适用于确定模仿人类情感的其他方法中的最佳方法（例如，手势、言语）。
Humans convey information about their emotional state through facial expressions. Robots typically cannot show facial expressions like humans do, making it hard for them to imitate emotions. Here we investigate how LED patterns around the eyes of Aldebaran's Nao robot can be used to imitate human emotions. We performed two experiments. In the first experiment we examined the LED color, intensity, frequency, sharpness, and orientation that humans associate with different emotions. Based on the results, 12 LED patterns were created. The second experiment measured how well humans recognized those LED patterns as the emotions intended by the design. We used a ROC (Receiver Operating Characteristic) graph to determine which of the 12 LED patterns were the best ones for the Nao robot to imitate emotions with. Our technique of using ROC graphs is generally applicable to determining the best of other methods for imitating human emotions (e.g., gestures, speech), as well.