Position-Based Multi-Agent Dynamics for Real-Time Crowd Simulation .

Exploiting the efficiency and stability of Position-Based Dynamics (PBD), weintroduce a novel crowd simulation method that runs at interactive rates forhundreds of thousands of agents. Our method enables the detailed modeling ofper-agent behavior in a Lagrangian formulation. We model short-range andlong-range collision avoidance to simulate both sparse and dense crowds. On theparticles representing agents, we formulate a set of positional constraintsthat can be readily integrated into a standard PBD solver. We augment thetentative particle motions with planning velocities to determine the preferredvelocities of agents, and project the positions onto the constraint manifold toeliminate colliding configurations. The local short-range interaction isrepresented with collision and frictional contact between agents, as in thediscrete simulation of granular materials. We incorporate a cohesion model formodeling collective behaviors and propose a new constraint for dealing withpotential future collisions. Our new method is suitable for use in interactivegames.