LCSim: A Large-Scale Controllable Traffic Simulator

2024-06-28

arXiv.org

AI Summary

Existing traffic simulators struggle to generate large-scale, high-fidelity urban scenarios with diverse driving styles, hindering robust evaluation of autonomous driving systems. To address this, we propose the first controllable traffic simulation framework based on diffusion models. Our method unifies multi-source traffic data into a coherent scene representation and introduces a conditional guidance mechanism to explicitly control driving styles—including aggressive, conservative, and stochastic behaviors. This approach overcomes the dual limitations of conventional rule-based and data-driven simulators in behavioral diversity and scene generalizability, enabling real-time, high-fidelity simulation over city-scale road networks with thousands of vehicles. Experiments demonstrate that our platform supports systematic manipulation of key variables—such as traffic density and style composition—thereby significantly enhancing the effectiveness and interpretability of autonomous driving algorithm evaluation, particularly in edge-case scenarios.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Diffusion Models for Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

With the rapid growth of urban transportation and the continuous progress in autonomous driving, a demand for robust benchmarking autonomous driving algorithms has emerged, calling for accurate modeling of large-scale urban traffic scenarios with diverse vehicle driving styles. Traditional traffic simulators, such as SUMO, often depend on hand-crafted scenarios and rule-based models, where vehicle actions are limited to speed adjustment and lane changes, making it difficult for them to create realistic traffic environments. In recent years, real-world traffic scenario datasets have been developed alongside advancements in autonomous driving, facilitating the rise of data-driven simulators and learning-based simulation methods. However, current data-driven simulators are often restricted to replicating the traffic scenarios and driving styles within the datasets they rely on, limiting their ability to model multi-style driving behaviors observed in the real world. We propose extit{LCSim}, a large-scale controllable traffic simulator. First, we define a unified data format for traffic scenarios and provide tools to construct them from multiple data sources, enabling large-scale traffic simulation. Furthermore, we integrate a diffusion-based vehicle motion planner into LCSim to facilitate realistic and diverse vehicle modeling. Under specific guidance, this allows for the creation of traffic scenarios that reflect various driving styles. Leveraging these features, LCSim can provide large-scale, realistic, and controllable virtual traffic environments. Codes and demos are available at https://tsinghua-fib-lab.github.io/LCSim.

Problem

• Modeling large-scale urban traffic scenarios.
• Creating realistic and diverse vehicle behaviors.
• Providing controllable virtual traffic environments.

Innovation

• Large-scale traffic simulation
• Diffusion-based vehicle motion
• Unified traffic data format