UrbanV2X: A Multisensory Vehicle-Infrastructure Dataset for Cooperative Navigation in Urban Areas

TL;DR

UrbanV2X presents a real-world multisensory dataset designed for vehicle–infrastructure cooperative navigation in dense urban settings. By integrating onboard cameras, LiDARs, 4D radar, UWB, IMU, GNSS-RTK/INS with roadside LiDAR, GNSS, and UWB and synchronizing everything with Precision Time Protocol, it enables robust cross-view perception and localization research. The dataset includes three urban scenarios, centimeter-level ground-truth poses, and a colorized LiDAR map generation pipeline, along with benchmarking of state-of-the-art localization methods. This resource addresses the critical need for infrastructure-aware datasets to advance V2X perception, cooperative SLAM, and smart city applications. Public availability accelerates reproducibility and progress in autonomous and connected mobility.

Abstract

Due to the limitations of a single autonomous vehicle, Cellular Vehicle-to-Everything (C-V2X) technology opens a new window for achieving fully autonomous driving through sensor information sharing. However, real-world datasets supporting vehicle-infrastructure cooperative navigation in complex urban environments remain rare. To address this gap, we present UrbanV2X, a comprehensive multisensory dataset collected from vehicles and roadside infrastructure in the Hong Kong C-V2X testbed, designed to support research on smart mobility applications in dense urban areas. Our onboard platform provides synchronized data from multiple industrial cameras, LiDARs, 4D radar, ultra-wideband (UWB), IMU, and high-precision GNSS-RTK/INS navigation systems. Meanwhile, our roadside infrastructure provides LiDAR, GNSS, and UWB measurements. The entire vehicle-infrastructure platform is synchronized using the Precision Time Protocol (PTP), with sensor calibration data provided. We also benchmark various navigation algorithms to evaluate the collected cooperative data. The dataset is publicly available at https://polyu-taslab.github.io/UrbanV2X/.

Figures (9)

• Figure 1: An overview of the setup of vehicle sensor-kit and roadside equipment, with visualization of the dataset. Top: the vehicle sensor-kit and roadside sensors setup. Each sensor is marked with a corresponding labeled box. Bottom: the visualization dataset, including multiple surround-view images, a sky-pointing image, point clouds from LiDAR and 4D radar, GNSS-RTK/INS and UWB data.
• Figure 2: System Architecture
• Figure 3: Sensor Frame
• Figure 4: The calibration between LiDAR and Camera. (a) Custom-Designed ArUco Calibration Board. ArUco markers are symmetrically attached around the rectangular calibration board. (b) The corners extracted from the image and point cloud in 3D space. (c) Extracting the 3D corner points of the calibration board from the image. The green parts represent the segmented calibration board point cloud plane and corner points, while the red parts represent the calibration board corner points calculated from the ArUco markers detected in the image. (d) Image fused with point cloud. The fusion effect is verified by projecting the point cloud onto the image.
• Figure 5: The extrinsics calibration for LiDAR and IMU. (a) LiDAR-IMU calibration. The image illustrates the process of sensor motion calibration using the LI-Init LiDARCalibrMars2022 method. The point cloud shows the constructed point cloud structure, while the blue line represents the motion trajectory. (b) LiDAR-LiDAR calibration. The image shows the fused point cloud after the registration of two LiDARs, where the white point cloud represents the Velodyne data, and the colored point cloud represents the Hesai data.