FLUID: A Fine-Grained Lightweight Urban Signalized-Intersection Dataset of Dense Conflict Trajectories

The FLUID dataset provides fine-grained trajectories with a unique focus on dense traffic conflicts at typical urban signalized intersections. Captured by drones, it contains approximately 5 hours of data covering over 20,000 traffic participants across three distinct intersection types. The dataset is distinguished by its high conflict rate (on average of 2.8 vehicle conflicts per minute across all scenes) and comes as a complete package with trajectories, traffic signals, maps, and raw videos, making it ideal for traffic behavior modeling and autonomous driving research.The Data descriptor has been accepted by Scientific Data. The title is “A Fine-Grained Lightweight Urban Signalized-Intersection Dataset of Dense Conflict Trajectories”.FLUID数据集（FLUID dataset）提供细粒度交通轨迹，其独特聚焦于典型城市信号交叉口的密集交通冲突场景。该数据集由无人机采集，包含约5小时的有效数据，覆盖三种不同的交叉口类型，涉及超过20000名交通参与者。本数据集的显著特色为高冲突率：所有场景下平均每分钟发生2.8起车辆冲突。其以完整配套套件形式对外发布，涵盖轨迹数据、交通信号信息、电子地图及原始视频素材，非常适用于交通行为建模与自动驾驶相关研究。本数据集的描述文稿已被《科学数据》（Scientific Data）期刊录用，论文标题为《细粒度轻量化城市信号交叉口密集冲突轨迹数据集》。