Modeling Of Communication Atmosfield In Human-Robot Interaction Based On Fuzzy Analytical Hierarchy Process And Its Application To Human-robot Interaction System

With the rapid development of artificial intelligence technology,intelligent robot has become a research hotspot in the field of emotion computing and human-robot interaction.People generally expect that robots have the ability to recognize and understand emotions and communication atmosphere like humans.The communication atmosphere is the psychological feeling of people during communication among humans and robots,which can not only reflect the overall communication,but also affect the communication.Therefore,in human-robot interaction,if the robot can not only recognize human's emotional state but also perceive the overall communication atmosphere,the robot can make appropriate behavior feedback and promote natural and harmonious human-robot interaction.For the communication atmosphere in human-robot interaction,the thesis has conducted a deep research.A fuzzy atmosfield model based on fuzzy analytical hierarchy process is proposed,which has been applied into a human-robot interaction system.The research of this thesis about communication atmosphere mainly includes three parts.Firstly,the fuzzy atmosfield model based on fuzzy analytical hierarchy process is proposed.The fuzzy atmosfield of human-robot interaction based on the emotional states of human,robot and background music is introduced,and the three-level structure of human-robot interaction is proposed.According to the real-time acoustic features of humans,robots and background music,including volume,pitch,rate,and duration,fuzzy analytic hierarchy process is used to calculate dynamic weights of the emotional states of humans,robots and background music on communication atmosphere.At the same time,the relationship between local communication atmosphere and global communication atmosphere is analyzed,and the calculation method of global atmosphere based on local atmosphere and emotional state of background music is constructed.Finally,the image representation method of fuzzy atmosfield based on shape,color,fill,and waveform is introduced,which can visualize the communication atmosphere.Secondly,a music emotion recognition model based on two-layer SVM is proposed.For different music clips,the musical genre is identified firstly,and the emotional state of the music clip is recognized based on musical genre.For the emotional recognition of music,features related to music genre and features related to music emotion should be extracted firstly.Then the SVC model is used to determine the musical genre using features related to music genre,and the corresponding SVR model is selected to recognize the musical emotion based on the recognition result of musical genre.Compared with recognizing musical emotion directly,musical emotion recognition based on musical genre can improve the recognition efficiency and accuracy.Finally,a human-robot interaction system based on proposed fuzzy atmosfield model is built.The fuzzy atmosfield model has been applied to the human-robot interaction system and the human-robot interaction interface is established.The system can collect emotional data of images and speech using Kinect,and use related algorithms to realize the recognition of emotional states and communication atmosphere.The robot can make appropriate behavior feedback according to human's emotional states and communication atmosphere.The scene experiments of communication atmosphere in a home environment are designed.The experiment system can reflect the real-time communication atmosphere of each scene experiment,and adjust communication atmosphere through background music,which can promote harmonious and smooth human-robot interaction.Compared with the subjective evaluation of experimenters using questionnaire survey,the experimental results show that the communication atmosphere obtained by the fuzzy atmosfield model is basically the same as humans' perception.