Towards Real-Time Crowd Simulation Under Uncertainty Using an Agent-Based Model and an Unscented Kalman Filter

Abstract

Agent-based modelling (ABM) is ideally suited to simulating crowds of people as it captures the complex behaviours and interactions between individuals that lead to the emergence of crowding. Currently, it is not possible to use ABM for real-time simulation due to the absence of established mechanisms for dynamically incorporating real-time data. This means that, although models are able to perform useful offline crowd simulations, they are unable to simulate the behaviours of crowds in real time. This paper begins to address this drawback by demonstrating how a data assimilation algorithm, the Unscented Kalman Filter (UKF), can be used to incorporate pseudo-real data into an agent-based model at run time. Experiments are conducted to test how well the algorithm works when a proportion of agents are tracked directly under varying levels of uncertainty. Notably, the experiments show that the behaviour of unobserved agents can be inferred from the behaviours of those that are observed. This has implications for modelling real crowds where full knowledge of all individuals will never be known. In presenting a new approach for creating real-time simulations of crowds, this paper has important implications for the management of various environments in global cities, from single buildings to larger structures such as transportation hubs, sports stadiums, through to entire city regions.