A Group Behavior Model of Real-time Crowds in Virtual Environments

We present a real-time crowded model in virtual environments. In our model, we simulate the group behavior of huge crowds in real-time. Considering the limitation of computing speed, we adopt a hierarchical management model and construct the movement area in virtual environment by a dynamical potential field. The improved leader-follower structure yields realistic group behavior comparatively. Combination of global planning and local collision avoidance in a optimize framework solves the problems in motion of large crowds without effective collision avoidance. The test results illuminates that the smooth motion of every individual can be simulated on interactive rate in a mainstream microcomputer.