Exploring Spatiotemporal Patterns of Pedestrian Movements in Shopping Streets using Agent-Based Modelling

Abstract

This report discusses the results of a masters thesis on the usefulness of dynamic modelling for simulation of pedestrian movement in shopping streets. The motivation for this research partly came from the introduction of social distancing measures in the Netherlands, due to the Covid-19 pandemic. Shopping streets are crowded locations for which it is likely that social distancing can be challenging, especially when obstacles or crowds block pedestrian flows.

Moreover, current studies on the topic of pedestrian movement patterns mainly focus on crowd management and situations such as evacuations and fastest or shortest route navigation, instead of movements under non-emergent and stress free conditions. However, research into pedestrian flows has practical value in a variety of other domains. Pedestrian movement simulation models can be useful for safety purposes, but also for the planning and design of public as well as private space, it is useful to know how people move in space. Dynamic spatial modelling allows spatial planners to acquire an idea of future conditions or possible effects of the plans or policies they are developing, prior to implementing them. This saves time and trials, and enhances consensus among stakeholders and formulation of appropriate proactive measures. There are multiple useful methodologies that can be used to get a better understanding of pedestrian movement patterns, such as cellular automata (CA) and agent-based modelling (ABM). While CA has successfully contributed to traffic flow studies, it is criticised for oversimplifications of reality. ABM is known for the rather microscopic level of modelling, in which the movement of individuals in complex systems can be analysed, while CA is more useful for macroscopic analyses. Therefore, in order to get insights into pedestrian movement patterns at street level, agent-based modelling software, in this case GAMA, has been used.

A conceptual model for the pedestrian simulation model has been developed to describe and simulate pedestrian movements. This study explored how this conceptual model could be translated to an agent-based model using the GAMA software. The presented model is used to demonstrate how the simulated pedestrian crowds are influenced by obstacles in a shopping street. Based on the observations and computational modelling experiments, it could be concluded that pedestrian movements in shopping streets can be successfully simulated with agent-based models, and that insights into movement patterns can be gained. However, for non-computer scientists, the GAMA software lacks easy-to-use tools to be able to model pedestrian movement accurately. Additionally, the model still needs to be calibrated and validated. To get better insights into the usefulness of the model, it is recommended for further research that such simulation models get calibrated with field observations. Then, dynamic and agent-based modelling allows urban planners, city authorities and decision makers to evaluate the impact of future urban design scenarios on pedestrian movements in shopping streets.