ABSTRACT
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Local navigation is one of the fundamental aspects of simulating crowds. Here, a crowd is viewed as any collection of many independent agents, each making its own navigation decisions. The task of a local navigation algorithm is to ensure that each agent in the crowd makes progress to its goals while smoothly and naturally avoiding collisions with others nearby. While agents are typically assumed to act in a decentralized fashion, there is generally an implicit goal that global human-like, high-level behavior patterns should emerge as the commutative result of the many local interactions between the agents.
