家庭服务机器人情感计算模型的研究  

Affective computing model based on HMM for home-service robot

作　　者：谢丽君[1] 文翰[1,2] 肖南峰[1] 

机构地区：[1]华南理工大学计算机科学与工程学院,广东广州510006 [2]佛山大学理学院,广东佛山528000

出　　处：《计算机工程与设计》2012年第1期322-327,共6页Computer Engineering and Design

基　　金：国家自然科学基金与中国民用航空总局联合基金项目(60776816);国家自然科学基金项目(61171141);广东省自然科学基金项目(8251064101000005)

摘　　要：根据基本情感理论建立了家庭服务机器人的情感状态概率空间模型,并应用马尔可夫链的特性,建立了基于隐马尔可夫模型的情感计算模型。详细地阐述了该情感计算模型中各参数的意义以及估算方法。通过仿真实验验证了该情感计算模型可以较好地模拟情感状态的自发转移,以及在外部刺激作用下的情感转移。通过对实验数据分析,发现机器人的情感经外部刺激作用或者自发演变,最终趋于稳定状态,这个稳定状态与情感转移概率矩阵有关,而与机器人所处的初始情感状态无关。According to the basic emotion theory, a probability space model of emotion states is built for home-service robot. By applying the characteristics of Markov chain, an affective computing model based on HMM is established. And the meaning and estimation methods of the parameters are elaborated in the affective computing model. The results of simulation experiment show this kind of affective computing model can better simulate the spontaneous transfer process of emotion states and the transfer process affected by external stimulus. In addition, the simulation experiments show that the emotion state of the robot would gradually become stable by spontaneous transfer process or the transfer process affected by external stimulus. The steady state is associated with emotional transition probability matrix and unrelated to the initial emotional state.

关 键 词：情感计算模型 隐马尔可夫模型 家庭服务机器人 情感状态 情感转移概率矩阵 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]