Reactive Human Motion Synthesis System

Over the recent years, the increasing demand for human animation in the fields of virtual reality, video games, and movie industry has greatly contributed to the development of computer animation technologies, especially the motion capture technology. Thanks to the motion capture technology, huge amount of motion data become available. Currently, the research focuses on:how to reuse the motion data to synthesize the reactive human motions which react faithfully to the external forces. Such kind of research is of great significance as the roles in such fields as games and virtual reality are required to react faithfully to the external forces. The traditional approach produced similar reactive motions based on physical simulation, however, motions produced in this way were somewhat distorted or not accurate enough, and they were not in line with the law of human motions.This paper aims to explore the reactive motions which simultaneously respond to the external forces on the basis of motion data. To begin with, by adopting physics simulation, this paper simulates a series of reactive motions of human when it is hit, then it sets up a dynamical model of human skeleton and relevant physical parameter, through physics simulation, the result of human motions in the physics world can be simulated; Secondly, a motion tracking controller is designed to actively control the motion simulation, track down the motion data of the target, and simultaneously react to the external forces. Finally, the thesis proposes a project to present a set of integrated human motions, in other words, from the motion database, it retrieves the follow-on motions which match the reactive motions simulated from the dynamical model, then it blends the above-mentioned motions into synthesized reactive motions.The contribution of this paper lies in:by integrating physical simulation and motion data, a reactive human motion synthesis system is designed which reacts accurately and simultaneously to the external forces under the premise preserving the authenticity of motion data.