Modeling crowd motion using swarm heuristics and predictive agents

We present a novel model of crowd motion and swarm behaviour, in the form of a multi-agent system. We introduce a set of behavioural rules arising from an individual’s presence within a swarm of other individuals. First, the separation force causes every agent to maintain a comfortable distance from all others in the same swarm. Second, the alignment force causes every agent in the same swarm to travel in the same direction as her nearest neighbour. Third, the cohesive force makes every individual in the same swarm remain in the vicinity of other swarm members. In addition to swarm forces, we provide each agent with the ability to predict its successive position and the successive position of those around it and to incorporate such information into the decision of its new position.