Natural discretization of pedestrian movement in continuous space

Area of Science:

• Computational Social Science
• Agent-Based Modeling
• Human Mobility Simulation

Background:

• Traditional cellular automata models for pedestrian movement are limited by their grid-based nature.
• Simulating realistic pedestrian dynamics requires models that allow for continuous movement in arbitrary directions.
• Empirical data on pedestrian step lengths and their relation to speed is crucial for model development.

Purpose of the Study:

• To develop a novel pedestrian movement model unbound by cellular grids, inspired by natural human walking.
• To incorporate empirically derived step lengths as model parameters.
• To validate the model's ability to reproduce known density-velocity relationships and simulate evacuation scenarios.

Main Methods:

• Developed a model using local discretization on circles around virtual pedestrians to allow continuous movement.
• Conducted a controlled experiment to collect empirical data on pedestrian step lengths relative to speed.
• Implemented a calibration algorithm to match density-velocity relations and reenacted an experimental evacuation scenario.

Main Results:

• The model successfully simulates pedestrian movement in arbitrary directions, overcoming grid limitations.
• Empirical data on step lengths was integrated, enhancing model realism.
• Simulated evacuation egress times closely matched experimental results, validating the model's predictive power.

Conclusions:

• The proposed model offers a flexible and empirically grounded approach to simulating pedestrian dynamics.
• The model's ability to reproduce known behaviors and experimental outcomes demonstrates its validity and potential applications.
• A new normalized measure for space occupancy aids in visualizing and analyzing simulation results.