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

2025-06-26

AI Summary

Existing urban traffic simulation platforms struggle to simultaneously achieve scalability, dynamic adaptability, and computational efficiency: conventional activity-based models require intensive data calibration; machine learning approaches suffer from poor generalizability; and LLM-based agent simulations incur prohibitive computational overhead. This project proposes a lightweight, city-scale dynamic mobility simulation platform featuring a novel “domain-specific generator + fine-tuned LLM” collaborative architecture, enabling batch generation of activity chains and context-aware dynamic revision. It integrates hybrid generative modeling with an environment-feedback-driven rescheduling mechanism, supporting concurrent validation of system-level and agent-level algorithms. The platform efficiently simulates full-day activity chains for 53,000 residents on a standard PC, responds in real time to disruptions (e.g., road closures, large-scale events), achieves over 3× higher computational efficiency than state-of-the-art baselines, and significantly improves behavioral fidelity.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Sampling/Simulation-based Search
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Web Mining and Content Analysis: Web data generation and simulation
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Understanding and modeling human mobility patterns is crucial for effective transportation planning and urban development. Despite significant advances in mobility research, there remains a critical gap in simulation platforms that allow for algorithm development, policy implementation, and comprehensive evaluation at scale. Traditional activity-based models require extensive data collection and manual calibration, machine learning approaches struggle with adaptation to dynamic conditions, and treding agent-based Large Language Models (LLMs) implementations face computational constraints with large-scale simulations. To address these challenges, we propose MobiVerse, a hybrid framework leverages the efficiency of lightweight domain-specific generator for generating base activity chains with the adaptability of LLMs for context-aware modifications. A case study was conducted in Westwood, Los Angeles, where we efficiently generated and dynamically adjusted schedules for the whole population of approximately 53,000 agents on a standard PC. Our experiments demonstrate that MobiVerse successfully enables agents to respond to environmental feedback, including road closures, large gathering events like football games, and congestion, through our hybrid framework. Its modular design facilitates testing various mobility algorithms at both transportation system and agent levels. Results show our approach maintains computational efficiency while enhancing behavioral realism. MobiVerse bridges the gap in mobility simulation by providing a customizable platform for mobility systems planning and operations with benchmark algorithms. Code and videos are available at https://github.com/ucla-mobility/MobiVerse.

Problem

• Scaling urban mobility simulation with hybrid lightweight and LLM models
• Overcoming data and computational limits in mobility pattern modeling
• Enabling dynamic agent responses to urban events and disruptions

Innovation

• Hybrid lightweight generator for base activities
• LLMs enable context-aware dynamic adjustments
• Modular design supports scalable mobility algorithms