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

TL;DR

MobileCity presents a scalable, tile-based urban mobility simulator that integrates a multi-modal transportation system with survey-derived agent profiles and a pre-generated action space to achieve realistic large-scale behavior. Its architecture—comprising profile, action generation, action selection, memory, and communication modules—enables efficient, human-like decision making and social interactions, aided by asynchronous memory updates and offline action generation. Empirical results on 4,000 agents show improved human-likeness, plausible venue usage, and demographic-consistent macro-patterns, while maintaining practical runtimes. The framework offers actionable insights for mobility analysis and urban planning, with open-source code to support adoption and further research.

Abstract

Generative agents offer promising capabilities for simulating realistic urban behaviors. However, existing methods oversimplify transportation choices, rely heavily on static agent profiles leading to behavioral homogenization, and inherit prohibitive computational costs. To address these limitations, we present MobileCity, a lightweight simulation platform designed to model realistic urban mobility with high computational efficiency. We introduce a comprehensive transportation system with multiple transport modes, and collect questionnaire data from respondents to construct agent profiles. To enable scalable simulation, agents perform action selection within a pre-generated action space and uses local models for efficient agent memory generation. Through extensive micro and macro-level evaluations on 4,000 agents, we demonstrate that MobileCity generates more realistic urban behaviors than baselines while maintaining computational efficiency. We further explore practical applications such as predicting movement patterns and analyzing demographic trends in transportation preferences. Our code is publicly available at https://github.com/Tony-Yip/MobileCity.

Figures (13)

• Figure 1: An agent traveling from apartment to company has three route options: (1) walking through zebra crossings (yellow lines), (2) walking to PMV (Personal Mobility Vehicle) station, riding on highway, then walking to destination (blue lines), or (3) walking to bus station, taking bus, then walking to destination (green lines).
• Figure 2: The crowd distribution across different urban venues, on weekdays (top) and weekends (bottom).
• Figure 3: Percentage point differences in activity distribution between our method and real-world data across demographic categories.
• Figure 4: Residents with different employment status have different fluctuations in basic needs during the day.
• Figure 5: The map of our simulated city.