Title: Agent-based simulation of pedestrian movement: a gradient method with an amplification parameter

Authors: Michal Dziecielski; Marcin Wozniak

Addresses: Faculty of Human Geography and Planning, Adam Mickiewicz University, Poznan, Poland ' Faculty of Human Geography and Planning, Adam Mickiewicz University, Poznan, Poland

Abstract: Pedestrian transportation is becoming increasingly important in modern cities. In this paper, we develop an agent-based model to simulate pedestrian traffic in a city. The real residential area represented by geographic information system (GIS) data was used. This paper aims to show the general model of pedestrian movements that can be applied to both small neighbourhood spaces and whole districts. It can be also used to simulate pedestrian traffic in open as well as closed spaces. A gradient method with an additional parameter related to the distance between the agent and the target was used. What were visualised in the first step were the trajectories of agents' movement and analysed lengths of routes, as well as the time needed to reach the destination by agents. The second step involved sensitivity analysis. Morris screening results show that the impact of all parameters is significant and highly diversified. Nonlinear dependencies between model inputs were detected. The article also shows different walking strategies, e.g., leisure walk or rush to the target. Simulations present how agents move to their destination by navigating and avoiding obstacles they face.

Keywords: agent-based modelling and simulation; pedestrian modelling; geographic information system; GIS; sensitivity analysis; Morris screening.

DOI: 10.1504/IJSPM.2022.126897

International Journal of Simulation and Process Modelling, 2022 Vol.18 No.3, pp.244 - 252

Received: 11 Mar 2022
Received in revised form: 20 Aug 2022
Accepted: 22 Aug 2022
Published online: 11 Nov 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article