Study On Service Robot Interactive Platform Based On Artificial Psychology

Affective and natural human-computer interaction is an emerging challenge in the field of computer applications which in recent years has been receiving attention of experts and scholars in the fields of psychology and computer science. In order to realize natural and intuitive human-computer interaction, they relate computer science with psychology and cognitive sciences, research on affective characters exposed during the interpersonal interaction and try to develop human-computer interaction systems with emotional feedbacks. Based on affective computation and artificial psychology, this thesis researches the technique and method that sets up the interactive platform of service robot. It discusses in detail control architecture of service robot, affective expression, affective modeling and the robot application system and offers technical supports for harmonious human-computer interaction. The main contents and creative points of this paper are as follows:1) Propose a kind of control architecture of service robot based on artificial psychology, give the control system layering structure based on the control flow and the state flow, and research the software and hardware structure of the robot platform. All these scenario form the system organization view from macro view point to micro's, which make the system more structural, more layered and logical, and draw a clear mission blueprint for the topic research. . In this architecture, the robot can obtain the cognition of environment through sensors, and after pretreatment and model treatment, and ultimately through its own acts to express the response to external stimulation. Meanwhile, the psychology of robot is inspired by the aim not only the environment, which can sufficiently show its active character and then generate behavior intent and give expression to it. To ensure the safety of robot platform, we focus on the"Three Laws of Robotics"engineering implementation issues and put forward a multi-sensor information fusion safety control method based on grey relation analysis and principal component analysis (PCA).The control architecture is verified to be realizable and practical on a robot prototype.2) In the robot behavior expression issue, the important function of bodily limbs posture and facial expression is described in a natural and harmonious interaction. In the motion of robot limbs, a motion planning method based on the third B spline curve is introduced. In Robot facial expressions, this paper firstly presented and realized the structural model of robot head, and adopted robot facial expression synthesis method based on action unit. The conception of affective state sets and affective expression sets are introduced, and a function from affective state to affective expression is set up, the affective function can be expressed by a set of productive rules. In order to make people feel that the robot is zoetic and friendly, face-to-face Human-Computer Interaction can be achieved through sound direction distinction. The proposed action expression methods are verified to be efficient on the affective robot prototype.3) Propose an emotional model for robot according to Maslow's hierarchy of needs and Siminov's mood information. Information that robot expect to receive and has obtained is multilevel and it is a multi-objective problem,so the analytic hierarchy process (AHP) is introduced. Using this method, we can build a personalized model according to designers'requirements to the robot character. In the actual use of emotion model, we have introduced a number of ancillary procedures to enable the robot to better simulation of human emotions. In order to make the robot know what it has done and express emotion in a better way, a learning model based on the probability of incidents is introduced for it. According to user's feedback, learning model would classify the robot act events set as positive events set, general events set, and negative incident sets. When the robot generated negative sentiment, in accordance with learning model, based on positive events set, it would change interactive topic, transfer users attention, make the interactive atmosphere head in the right direction.4) Based on the researches mentioned above, a hominine robot is fully realized. This robot has hominine shape, bodily limbs and facial expression, so can carry out multimodal interaction with human. Nowadays, our research of affective robot has reached the lead level domestically. Experiments show that robot can interact with human in a natural and harmonious way.