Emotional communication between human and machine using motion of joints of body

这项研究的目的是实现人与机器人之间的情感交流。学者们曾多次尝试证明，粪便和声音的表达在接受和表达人类和机器人之间的情感方面是有效的。我们用身体动作来表达情绪，而不是排泄物和声音。我们取得了以下成果：建立了一个用于情感研究的电影数据库。运动捕捉系统被用来获取人体步态信息。沃克的主要关节上安装了12个传感器。演员们被要求在走路时表达情感。该数据库包括30名演员的步态信息。我们从步态显示中调查了人类对情感的感知。步态数据以生物运动的形式呈现。人类受试者被要求回答从展示中感受到的情绪。正确率为40%。我们构建了一个情感识别系统。研究了一种从人体步态模式的感知数据中提取人体情感的特征提取方法。我们假设高维生物运动数据是由统计上相互独立的低维特征生成的。通过对给定的生物运动数据的判别结果来评估提取的特征，所述生物运动数据识别出五种类型的类别，即：愤怒、悲伤、厌恶、喜悦和恐惧。利用3个参与者的生物运动数据，对5类情绪进行识别，正确率达到40%。

The purpose of this research was to achieve emotional communication between humans and robots. Numerous attempts have been made by scholars to show that the expressions of feces and voices are effective in receiving and manifesting emotion between humans and robots. Instead of feces and voices, we use body movement as a cue to emotion. We have obtained following results.We have built a movie database for emotion study. A motion-capturing system was used to get human gait information. 12 sensors were put onto walker's principal joints. Actors were asked to express emotion while walking. The database includes gait information from 30 actors.We investigated human percept on emotion from gait display. The gait data were presented as biological motion. Human subjects were asked to answer perceived emotion from the display. The rate of correct answer was 40%.We constructed an emotion-recognition system. A method of feature extraction to discriminate emotions of human from a sensing data of human gait patterns was studied. We assume that the high dimensional biological motion data are generated by low dimensional features which components are statistically independent. The extracted feature is evaluated by a discriminated result of the given biological motion data which identified five types of categories, "Anger", "Grief", "Disgust", "Joy" and "Fear". We achieve 40% accuracy for 5-class of emotion discrimination with 3 actors' biological motion data.