Research On Human-Computer Interaction Emotional Response And Guidance Methods

With the rapid development of human-computer interaction technology,more and more attention has been paid to the influence of the robot’s emotional state on the entire interaction process,focusing on the robot’s emotional response process and the emotional guidance process.The aim is to enhance the robot’s affective computing capabilities in the interaction process,giving the robot unique and active characteristics that enable it to respond to participant input and guide the participant’s emotions towards the target emotion.This thesis first introduces Affective Computing(AC)for robots,and then constructs different models and validates the effectiveness of the proposed models to address the problems of achieving robot affective response and guiding participant affect.The main work and innovations are as follows:This thesis addresses the problem that robots cannot express unique,natural and vivid emotions in existing human-computer interaction systems,which leads to low participant satisfaction and experience.An emotional response model for robots based on the Pleasure-Arousal-Dominance(PAD)emotion space is proposed,which based on a fuzzy cognitive map.The process of robot emotional response in human-computer interaction is modelled by considering giving the robot human-like personality traits and social roles.By obtaining participant emotion values based on the context of the interaction,robots that incorporate personality traits and social roles develop a unique sense of self and are able to take advantage of the moment to actively engage in the conversation.The model increases the uniqueness and proactivity of the robot’s emotional response to the participant’s interaction and influences the participant’s input in the next moment.The experimental results show that the proposed model can increase the uniqueness and uniqueness of the robot’s emotional response and effectively improve the participants’ interaction satisfaction and experience.The problem of the lack of initiative and guidance in the robot’s emotional expressions during human-computer interaction and the tendency to cause ups and downs in the participant’s emotional experience are addressed.In this study,the trend of emotion change in continuous time is considered,and a reinforcement learning method is used to guide the participant’s emotion state to the target emotion state and maintain it near the target emotion state by increasing the robot’s initiative of emotion expression.Firstly,the participant’s emotion change curve for a known target emotion state is plotted based on the statistical properties of emotion.Secondly,a progressive emotional state of identity,security and belonging is used as the reward function for reinforcement learning,taking into account the psychology of social interaction and the participant’s interaction needs.Finally,the method uses reinforcement learning to build an emotional guidance model for intelligent robots with a feedback mechanism from the participant’s interaction experience and emotional accuracy,so that the robot can guide the participant to reach the target emotional state in the process of continuous interaction.The experimental results show that the proposed method can gradually guide the participant to the target emotion while ensuring the emotional accuracy,and further enhance the participant’s interaction satisfaction.