Crowd simulation: A video observation and agent-based modelling approach

Human movement in a crowd can be considered as complex and unpredictable, and accordingly large scale video observation studies based on a conceptual behaviour framework were used to characterise individual movements and behaviours. The conceptual behaviours were Free Movement (Moving Through and Move-Stop-Move), Same Direction Movement (Queuing and Competitive) and Opposite Direction Movement (Avoiding and Passing Through). Movement in crowds was modelled and simulated using an agent-based method using the gaming software Dark BASIC Professional. The agents (individuals) were given parameters of personal objective, visual perception, speed of movement, personal space and avoidance angle or distance within different crowd densities. Two case studies including a multi-mode transportation system layout and a bottleneck / non-bottleneck evacuation are presented.