Profiling Research On The Face Transforming Mechanism Of Human-head Portrait Robot

With the development of science and technology, robot technology is moreemotional, intelligent and personalized. Robots participating in human’s life andlearning and robots’ harmonious coexistence with human beings are the goals of robotresearch. Human-head portrait robot has similar human face shape and functions withhuman and can help the emotion and information communication between robots andhuman beings. Currently, human-head portrait robot has become a hot research topic inthe robot area. However, robot with one image cannot show the variety of human faces,in order to enhance the effect of the robot to interact with people, the research on facetransforming mechanism of human-head portrait robot is proposed in the paper.This paper defined the feature points that played decisive role on the shape of faceby analyzing the characteristics of the face geometry, and established thetwo-dimensional mathematical model of facial contours by piecewise smooth curve.The head contour obtained from100real human head-face images verified theeffectiveness and universality of the model. The shape of face was divided into roundface, square face, egg-shaped face, heart-shaped face and long face according tohuman’s knowledge and experience. Five kinds of facial contours were obtained and thestandard parameters of the faces were calculated through cognitive experiments. Furtherclassification method was proposed based on the mathematical model of facial contoursand the range of parameters for different kind of face was given.The defined points were distributed uniformly as driving points of the mechanismto achieve deformation by considering facial skin’s characteristics. The facetransforming mechanism which had10driving points and12degree of freedom wasdriven by3main drivers and3groups of control components led by vice of screw andnut and sliding gear. The control progress of the mechanism was designed.First, the rigid model of virtual prototype which contained face transformingmechanism and facial skin was established in Solidworks. Then the facial skin was dealtwith flexible treatment in Ansys. Finally the simulation of50face samples including5kinds of face was completed in Adams. The results showed that the similarity of virtualprototype and the sample face facial contours was above82%, and the profiling errorwas below4%. It verified the feasibility of face transforming mechanism and theselection of facial skin and driving points. All the work above provided theoretical basisand simulation-based verification for the development of prototype.