Multimodal sensor dataset from vehicle-mounted mobile mapping system for comprehensive urban scenes

Lu, Shengyu; Bao, Sheng; Shi, Wenzhong; Wei, Yitao; Zhang, Shuyu; Yang, Daping · 2025 · DOAJ

DOI: 10.1038/s41597-025-05471-1

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper introduces MSD-VMMS-HK, a comprehensive multimodal sensor dataset collected from a vehicle-mounted mobile mapping system in Hong Kong. The research addresses a critical gap in the mobile mapping field: existing datasets predominantly rely on portable platforms like backpacks or robots, which lack the scale and sensor richness required for large-scale urban mapping and autonomous driving research. Furthermore, current vehicle-mounted datasets often fail to cover challenging "non-exposed" environments, such as tunnels and urban canyons, where Global Navigation Satellite System (GNSS) signals are degraded or absent. To facilitate research in high-definition mapping, digital twins, and localization, this study provides a high-precision, large-scale dataset featuring diverse urban scenarios. The data was acquired using a custom vehicle-mounted platform equipped with cutting-edge sensors: a 128-channel Velodyne VLS-128 LiDAR, a high-accuracy single-line Riegl VUX-1HA LiDAR, a Ladybug5+ panoramic camera, and a NovAtel SPAN CPT7 GNSS/INS system. All sensors underwent rigorous geometric calibration and time synchronization using Pulse Per Second (PPS) signals to ensure nanosecond-level precision, preventing alignment errors during data fusion. The dataset comprises 2 TB of data collected across various Hong Kong locations, including mountain and cross-harbour tunnels, urban canyons, indoor parking lots, and highways. Notably, the Central District was recorded at three different times to enable change detection analysis. To establish ground truth in GNSS-denied areas, the authors employed a GNSS + INS + Simultaneous Localization and Mapping (SLAM) approach, optimizing odometry via factor graphs to maintain accuracy in enclosed spaces. Technical validation demonstrates the dataset’s high precision. Using government-provided control points, the point cloud data for Fat Kwong Street achieved a mean horizontal positioning error of 0.329 meters and a mean vertical error of 0.344 meters. In the more challenging How Ming Street scenario, the mean horizontal error was 0.629 meters, while the vertical error was 2.793 meters. The dataset provides synchronized point clouds, panoramic images, and IMU/GNSS data, offering a robust benchmark for evaluating mapping accuracy in complex environments. The significance of MSD-VMMS-HK lies in its provision of the first urban-level comprehensive dataset that includes high-precision references for non-exposed spaces. By addressing the scarcity of high-quality vehicle-mounted data in GNSS-challenged environments, it supports advancements in SLAM, localization, and autonomous driving. The inclusion of multi-temporal data and diverse scenarios allows researchers to validate algorithms for urban infrastructure management, change detection, and accurate mapping in conditions where traditional GNSS-based methods fail.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• v2x connected vehicle

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Methodological Resource: dataset resource