# MobileCity: An Efficient Framework for Large-Scale Urban Behavior Simulation

**Authors:** Xiaotong Ye, Nicolas Bougie, Toshihiko Yamasaki, Narimasa Watanabe  
**arXiv:** [2504.16946](https://arxiv.org/abs/2504.16946)  
**Date:** 2025-04-18  
**Summary page:** https://www.summarizepaper.com/en/paper/2504.16946v1/

## Key points

- Authors explore the potential of generative agents in simulating realistic urban behaviors
- Existing methods oversimplify transportation choices and require extensive computational resources for large-scale population simulation
- Introduce a virtual city with multiple functional buildings and transportation modes
- Conduct thorough surveys to understand behavioral choices and mobility preferences among different population groups
- Develop a simulation framework that captures the complexity of urban mobility while remaining scalable enough to simulate over 4,000 agents
- Validate realism through detailed micro and macro-level analyses
- Perform experiments such as predicting crowd density based on movement patterns and identifying trends in vehicle preferences across various agent demographics
- MobileCity offers an efficient framework balancing realism with scalability, advancing urban behavior simulation and contributing valuable insights into urban dynamics.

## Summary

In their paper titled "MobileCity: An Efficient Framework for Large-Scale Urban Behavior Simulation," authors Xiaotong Ye, Nicolas Bougie, Toshihiko Yamasaki, and Narimasa Watanabe explore the potential of generative agents in simulating realistic urban behaviors. They highlight the limitations of existing methods that oversimplify transportation choices in modern cities and require extensive computational resources for large-scale population simulation. To address these challenges, the authors introduce a virtual city with multiple functional buildings and transportation modes. They conduct thorough surveys to understand behavioral choices and mobility preferences among different population groups. Leveraging these insights, they develop a simulation framework that captures the complexity of urban mobility while remaining scalable enough to simulate over 4,000 agents. To validate the realism of the generated behaviors, the authors perform detailed micro and macro-level analyses. Going beyond mere performance comparison, they delve into insightful experiments such as predicting crowd density based on movement patterns and identifying trends in vehicle preferences across various agent demographics. Overall, MobileCity represents a significant advancement in urban behavior simulation by offering an efficient framework that balances realism with scalability. The research contributes valuable insights into understanding urban dynamics and can potentially inform decision-making processes related to urban planning and transportation management.