An Agent Based Model for London: Exploratory Case Study

Abstract

Agent based modelling is a modelling paradigm which may transform how key policy decisions are made. Agent Based Models (ABMs) simulate actions and interactions of agents with a view to assessing their effects on the system as a whole. They aim to provide insight into complex behaviour and simulate collective behaviour of agents, often by obeying simple rules, typically in natural systems. Their specific use in transportation and city scale modelling is documented in research but has only recently being looked at for real-world applications and decision-making due to a range of reasons, such as a lack of available data and computational infrastructure. Questions such as "What are the equity impacts of this transport scheme?" or "Will my policy decision have a detrimental impact on those who have to travel?" or "How does this decision affect the network outside the traditional peaks?" are difficult to answer with existing modelling tools. Furthermore, future mobility modes, such as Connected and Autonomous Vehicles, and frameworks, such as Mobility as a Service, pose further challenges to existing modelling tools. ABMs lend themselves well to tackling these questions and scenarios because of the agent level interaction, heuristic decision making, continuous modelling, and the vast socio-demographic data that can sit behind each agent (individual). In late 2018, Arup and TfL engaged in a collaborative effort to explore the development of a proof of concept ABM for London. This was based on previous work Arup carried out for Transport Victoria in Australia, where the team built a city-wide ABM of Melbourne using an open source framework, MATSim and on TfL’s work on population synthesis and activity based plan estimation. The joint team developed a synthetic population and built a UK-wide network to simulate travel over 24 hours in London. One of the key objectives of the collaboration was knowledge-share as well as learning. This paper sets out the fundamental principles behind ABMs, their potential application in answering difficult policy questions and the methodology, learnings and outcomes from the Arup/TfL collaboration.

Gerry Casey

Principal Research Fellow, UCL & Associate, Arup

Transport modeller and data scientist building city-scale simulations to help governments and cities make better decisions on transport, climate, and equity.