Research Of The Key Technologies For Autonomous Virtual Human

The technology of autonomous virtual human simulates the figure, motion, behavior and cognitive style of human beings in virtual environment. The research aim is to enable the characters to choose and control own behaviors according to the internal and external conditions without user intervention. With more and more virtual characters appearing in the applications of virtual training, industry designing and entertainment, people bring up higher requirements to the intelligence and autonomy of virtual human.Therefore, the autonomous virtual human has become a hotspot in the research field of virtual human. To solving the problems in the control of autonomous virtual human, this paper studied on some key technologies, such as motion control, perception, path plan and behavior control.Fistly, this paper studied on the control method of limbs motion of virtual human. The manipulability matrix from neuroscience was employed in the process, and combined with dynamic interpolation and optimal solution of IK to obtain actions according with the motion rules of human limbs. In multi-obstacle enviorment, this control method can generate conlision-free animation in real time.Secondly, the perception model of virtual human was established by analyzing the physiological characters and perceptive behaviors of human's sense organs. The model provides the perceptive faculties of vision, audition, touch and memory. While simulated the perceptive limitation of human beings, this model also realized the feedback influence of results on perceptive process.According to the specialties of the planning problem of virtual human, this paper integrated the arithmetic of global and local planning to establish the path planning model based on mixing information, which can obtain an optimum path according to the known condition in memory, and then evade dynamic or unknown obstacles by adjusting moving direction and velocity in real time.An autonomous behavior system was established in the paper, which realized the autonomous behavior choosing and decomposing from abstract behaviors to motion and action streams. Then a mixing behavior control archtechture was presented. Based on this archtechture, the perception, planning, motion and behavior systems was intergrated to establish a whole behavior control model of autonomous virtual human. Finally, this paper applied the behavior control model into industrial design field, and developed an ergonomics system based on virtual human technology. The system realized the evaluation of accessibility and visibility of productions or working spaces, and animation of evaluating process can be generated. By applied into this system, the behavior control model was validated to be extensible and flexible.