A Potential Based Many-Particle Model for Pedestrian Flow

We propose a many-particle model for pedestrian dynamics, which simply assumes that the velocity (including both magnitude and direction) of a pedestrian depends on the positions of all pedestrians. This is similar to that in the well-known social force model. However, our formulation is computationally efficient because the influences of all other pedestrians on the referred pedestrian are taken into account through the reconstruction of densities, the speed-density relationship, and the cost potential field that is also determined by densities. In the meantime, the formulation is able to simulate complex phenomena, such as the arching in front of exits, the formation of lanes in counterflow.