Modeling Virtual Human And Studying Its Swimming Motion Control

With the development of Computer Application Technology, Computer Animation, Biology Dynamics, Mechanics, Artificial Intelligence and other related disciplines, the range of research on virtual human behavior become more and more wide, which includes apperceive model of human language, modeling of virtual circumstance, computer aided education, information and communication technology, military simulation and so on. Virtual human has been penetrating into all aspects of our daily life. The research on virtual human focuses on the geometry modeling and behavior modeling. In the paper, we research the technology of swimming motion control of virtual human, which includes two parts: one is geometry modeling of virtual human, the other is the behavior modeling of swimming of virtual human.Now the principal methods of geometry modeling of virtual human includes bar model, face model, shape model and multi-hierarchy model. Since the highlight of the paper is to study the motion control technologies and the action of virtual human is mainly driven by the skeleton, we just take into account the layer of skeleton. For modeling, the structure of virtual human is simplified to the connectors of 23 bones and 23 joints, which is linking into tree structure. Parents and children of the bones are joints, and each bone can has zero or more than one child, but only one parent. Parents and children of joints are bones, and each joint can has only one child and one parent. For the bones could not deform in the locomotion, so we take them as multi-rigid systems. The aim of the thesis is to study the swimming motion of virtual human which is a cyclical movement. According to the movement of arms, the swimming motion includes three periods which are durations of pulling water, pushing water, and swinging arm. According to the idea of motional superposition, the swimming cycle is divided into four stages which are modeled respectively, by which the key frames of swimming motion are created. In the process of solving key frames, CCD(Cyclic-Coordinate Descent)algorithm is used to solve the inverse kinematics problems. And then, the spherally linear interpolation based on four-element vector is used to get the medial frames. The swimming track of the virtual human is set by hand. The trajectory of pelvis at the corner is dealt with particularly by using the Bezier curve. In addition, the introduction of some character parameters, such as step length, step frequency and so on, makes the motion diversely. At last, the running result of system shows that based on the methods above, we could get the fluent swimming motions.