LCSim: A Large-Scale Controllable Traffic Simulator

TL;DR

LCSim tackles the need for large-scale, controllable urban traffic simulation by unifying scenario data from multiple sources into a Protobuf-based format and integrating a diffusion-based motion planner conditioned on rich scene context. The core system combines a scene encoder, a diffusion decoder, and guided loss functions to generate controllable, realistic motion plans for $T_f=8$ seconds, enabling multi-style vehicle behaviors. An RL-friendly Gym-like interface supports benchmarking and training agents in diverse driving styles, with experiments showing improved traffic-rule adherence and style controllability, plus city-scale scenario construction from real-world trajectories. This approach enables scalable benchmark environments for autonomous driving algorithms and RL policies, supporting both realism and controllability in large, diverse traffic scenarios.

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 \textit{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.

Figures (12)

• Figure 1: The unified format of traffic scenario data.
• Figure 2: Traffic scenarios from different data sources.
• Figure 3: The simulation architecture of LCSim.
• Figure 4: The architecture of diffusion decoder.
• Figure 5: The process of generating vehicle motion plans by diffusion model.