Modeling pedestrian behaviors under attracting incidents using cellular automata

• Pedestrians are classified into three groups when incidents occur.
• Behaviors of the three styles of pedestrians are analyzed.
• A new cellular automaton model considering incidents is proposed.
• This model can capture the crowd macro-characteristics when incidents occur.

Compared to vehicular flow, pedestrian flow is more complicated as it is free from the restriction of the lane and more flexible. Due to the lack of modeling pedestrian behaviors under attracting incidents (incidents which attract pedestrians around to gather), this paper proposes a new cellular automata model aiming to reproduce the behaviors induced by such attracting incidents. When attracting incidents occur, the proposed model will classify pedestrians around the incidents into three groups: the “unaffected” type, the “stopped” type and the “onlooking” type. The “unaffected” type represents the pedestrians who are not interested in the attracting incidents and its dynamics are the same as that under normal circumstances which are the main target in the previous works. The “stopped” type represents the pedestrians are somewhat interested in the attracting incidents, but unwilling to move close to the venues. Its dynamics are determined by “stopped” utility which can make the pedestrians stop for a while. The “onlooking” type represents the pedestrians who show strong interest in the attracting incidents and intend to move close to the venues to gain more information. The “onlooking” pedestrians will take a series of reactions to attracting incidents, such as approaching to the venues, stopping and watching the attracting incidents, leaving the venues, which have all been considered in the proposed model. The simulation results demonstrate that the proposed model can capture the macro-characteristics of pedestrian traffic flow under normal circumstances and possesses the fundamental characteristics of the pedestrian behaviors under attracting incidents around which a torus-shaped crowd is typically formed.