Human-Robot Interaction And Cooperation With Humanoid Robot

The perception smart times promote artificial intelligence into the combination of computing technology with psychology, brain science, cognitive science and neuroscience. During the process of intelligent human-computer interaction, humanoid robot not only possesses humanoid shape, but also need the more important things, such as humanoid affection, humanoid thought, humanoid behavior, to meet the social service demands of the elder and disabled, spiritual consolation, rehabilitation, and so on. This dissertation mainly researched the affective-decision making and associative memory process during face interaction with a humanoid robot. Firstly, the Benchmark test course that was of artificial emotion was finished by means of spontaneous emotional transference and stimulus transference based on the experimental psychology paradigm with respect to children’s gambling task. Secondly, the multi-Agent affective-decision based on the affective ratiocination was constructed by introducing QSIM reasoning and AHP into multi-agent system, and the complex decision-making for the humanoid robot in the smart home environment was solved by virtue of the hierarchical structured approach. Thirdly, referencing to emotional energy theory, the adjustment algorithm about mood state caused by the combination of spontaneous transference and stimulus transference upon HMM was established, of which these jobs finally realized associative memory model of a humanoid robot. Finally, the theory and algorithm mentioned above were integrated into the platform of man-machine interaction facial expression, and the effectiveness of the model was analyzed and tested.The main innovation points are as follows:(1) The paradigm’s descriptive model was built based on the probabilistic finite-state machine, which was with respect to the research of Benchmark problem with relation to the engine produced by simulated emotion under psychological test paradigm. The children’s gambling task was also taken as a part of Benchmark, and the universality and effectiveness were used as evaluation index to evaluate the performance of engine from four aspects which were the variation of emotional intensity, playing cards, earnings and choice times. Under theoretical framework of Markov chain, the testing process was investigated by virtue of emotion generation model with spontaneous and stimulus transference. The experimental results show that emotion generation model can simulate the psychological process more realistically in the children’s gambling task, which can provide uniform input foundation for the psychology model validation.(2) According to the effective decision making architecture, the interior of an agent could be individed into four modules, decision-making analysis module, user interaction module, data processing module and information routing module, under the basis of the traditional multi-agent group decision-making system framework. Therefore, the actual data acquisition, decision-making objectives and mutual collaboration capabilities could be realized via operational utilization of responsiveness, autonomy and sociality, which provided the guarantee for the harmonious interaction of man-machine. In addition, the QISM reasoning and AHP were introduced into the multi-agent affective decision-making system, and the problem existing in the complex decision-making were solved by virtue of hierarchical structured approach, which further realized the reasoning function with respect to network topology, environment State, affective forecasting, and robot fault. Thus, the multi-agent emotional decision-making system upon affective reasoning machine was established in the intelligent home environment so as to verify and optimize decision-making capability of a humanoid robot.(3) According to the theory of emotional energy, personalized expressive model was divided into the emotional transfer stochastic process with external stimulus, the spontaneous emotional transfer process, the stimulating mood transfer process and the spontaneous mood dynamic equilibrium control process. On this basis, we proposed a regulation algorithm combined with HMM spontaneous transfer and stimulation based on the state of mind. The associative memory model used for a humanoid robot was submitted based on the BP network. The mood spontaneous and stimulus transference was combined based on the HMM model, which finally made the humanoid robot possess dynamic mood variation and regulation capability.(4) Three aspects as the system architecture, hardware organization and function modules described the overall humanoid robot with14degrees of freedom. Based on emotional energy theory and its own characteristics, PAD (Pleasure-Arousal-Dominance) three-dimensional emotional space, emotional state transfer model, the intensity attenuation of emotion were described respectively. Humanoid robot control and emotional regulaiton process were implemented based on the expression of mental energy. The human-robot interaction was built on the basis of C#programming language and SQL Server database management system for realize the models, algorithms and interactive managemet proposed on this dissertation. Eventually, the emotional evaluation, emotional decisions, associative memory, emotional adjustment were intergrate into the human-computer interaction and cooperation process on this robot platform.