by Craig Reynolds
Crowds and other flock-like group motion are often modeled as interacting particle systems. These multi-agent simulations are computationally expensive because each agent must consider all of the others, if only to identify its neighbors. For large crowds, simple implementations are too slow since computation grows as the square of agent population. Faster approaches often rely on spatial hashing where a partitioning of space is used to accelerate crowd simulation. This same partitioning can form the basis of a scalable multi-processor approach to large, fast crowd simulations, as in [Quinn et al. 2003]. This paper describes an implementation of that approach for PLAYSTATION®3 which supports simulation and display of simple crowds of up to 15,000 individuals at 60 frames per second.
crowd simulation, multi-agent simulation, interacting particle systems, flocking, boids, behavioral animation, parallel processing, distributed processing, multi-processor.
Reynolds, C. (2006) Big Fast Crowds on PS3. In Proceedings of Sandbox (an ACM Video Games Symposium), Boston, Massachusetts, July 2006, pages ?-?.
Full paper (PDF, 0.8 MB)
Videos and lecture slides
Send comments to Craig Reynolds
Last update: October 6, 2006