Stochastic transition model for pedestrian dynamics

TL;DR

This paper introduces an extended stochastic cellular automata model for pedestrian dynamics that improves upon existing methods by incorporating path planning and interaction components, accurately reproducing fundamental flow characteristics and self-organizing behaviors.

Contribution

It presents a novel stochastic model combining cellular automata with path planning and interaction, enhancing realism in pedestrian movement simulations.

Findings

Successfully reproduces fundamental diagram shape

Captures self-organizing behavior in pedestrian flow

Handles uncertainties in human motion patterns

Abstract

The proposed stochastic model for pedestrian dynamics is based on existing approaches using cellular automata, combined with substantial extensions, to compensate the deficiencies resulting of the discrete grid structure. This agent motion model is extended by both a grid-based path planning and mid-range agent interaction component. The stochastic model proves its capabilities for a quantitative reproduction of the characteristic shape of the common fundamental diagram of pedestrian dynamics. Moreover, effects of self-organizing behavior are successfully reproduced. The stochastic cellular automata approach is found to be adequate with respect to uncertainties in human motion patterns, a feature previously held by artificial noise terms alone.