Urban Mobility Simulation Developments

The field of urban mobility simulation is undergoing significant advancements with the integration of large language models, generative agents, and hybrid frameworks. These innovations enable more realistic and dynamic simulations, allowing for better transportation planning and urban development. Researchers are focusing on creating scalable, flexible, and customizable platforms that can simulate complex urban behaviors and phenomena. Notable developments include the use of recursive value-driven approaches, lifelong learning mechanisms, and transformer models to simulate traffic and agent behavior. Some noteworthy papers in this area include:

  • MobiVerse, which proposes a hybrid framework for scaling urban mobility simulation.
  • CitySim, which models urban behaviors and city dynamics with large-scale LLM-driven agent simulation.
  • SceneDiffuser++, which presents a generative world model for city-scale traffic simulation.
  • GATSim, which introduces a novel framework for urban mobility simulation with generative agents.
  • InfGen, which proposes a scenario generation framework as next token group prediction.

Sources

MobiVerse: Scaling Urban Mobility Simulation with Hybrid Lightweight Domain-Specific Generator and Large Language Models

CitySim: Modeling Urban Behaviors and City Dynamics with Large-Scale LLM-Driven Agent Simulation

SceneDiffuser++: City-Scale Traffic Simulation via a Generative World Model

GATSim: Urban Mobility Simulation with Generative Agents

InfGen: Scenario Generation as Next Token Group Prediction

Built with on top of