A real-time control model for intelligent virtual human

Intelligent virtual human is widely required in educational, entertainment, and simulation software. Providing real-time motion control architecture is very important in design this kind of digital actor. This paper presents a three-level control strategy, with which virtual human can independently navigate in virtual world and quickly make response to their environment. This model improves the subsumption architecture by adding the ability of global planning. Collision avoidance and path planning are realized in the 0-level layer and the 1-level layer, respectively. Then a simple expert system in the 2-level layer takes the responsibility for task planning to realize more intelligent behavior. Finally, we developed a real-time human animation system-smart human system, in which this control model is applied for performing intelligent behavior of the digital actors.