Overcoming computational complexity: a scalable agent-based model of traffic activity using FLAME-GPU

Agent-based modelling (ABM) has significantly advanced the simulation of complex systems across various disciplines, including economic markets, traffic systems, and ecological research. By representing individuals as autonomous agents operating within a defined environment, ABMs facilitate the exploration of system behaviours and the testing of interventions in a controlled setting. However, the computational demands of ABMs, particularly in simulating large-scale systems, pose significant challenges. This article addresses these challenges by introducing a GPU-accelerated transport model for the Isle of Wight, leveraging the FLAME-GPU framework to enhance scalability and efficiency in traffic simulation. By comparing the proposed model with traditional, non-GPU accelerated simulations, such as SUMO, the study demonstrates improved performance metrics, including simulation speed and the ability to handle larger vehicle populations effectively. This contribution not only showcases the potential of GPU acceleration in overcoming computational constraints of ABMs but also provides a practical framework for simulating transport systems at a more granular and extensive scale than previously possible. Through detailed experimentation, the article illustrates the model’s capability to realistically simulate the vehicle population of the Isle of Wight, aiming for a balance between computational tractability and the accurate representation of complex traffic interactions.