Hilti-Trimble-Oxford Dataset
收藏资源简介:
Hilti-Trimble-Oxford Dataset 是由喜利得、天宝及牛津大学等机构联合构建的高质量视觉-惯性基准数据集,专为评估SLAM和基于平面图的定位算法而设计。该数据集包含30个序列,采集自瑞士同一建筑工地长达八个月的建设周期,涵盖了七层楼的空间范围,数据形式包括360度全景视频、同步IMU测量值以及二维平面图,并提供了通过高精度激光雷达-惯性SLAM系统获取的真实轨迹作为基准。数据采集采用消费级360相机与嵌入式IMU,旨在以低成本方案捕捉现实建筑环境中的动态光照、快速运动、重复结构等挑战性场景。该数据集主要应用于建筑进度监控、机器人自主导航及计算机视觉领域,致力于推动视觉SLAM和平面图参照定位技术在复杂工业环境中的实用化发展。
The Hilti-Trimble-Oxford Dataset is a high-quality visual-inertial benchmark dataset jointly constructed by institutions including Hilti, Trimble, University of Oxford and others, specifically designed for evaluating SLAM and floor-plan-based localization algorithms. This dataset comprises 30 sequences collected over an 8-month construction period at the same construction site in Switzerland, spanning a 7-floor spatial area. Its data modalities include 360-degree panoramic videos, synchronized IMU measurements, and 2D floor plans, and it also provides ground-truth trajectories obtained via a high-precision LiDAR-inertial SLAM system as the benchmark reference. The data was collected using consumer-grade 360 cameras and embedded IMUs, aiming to capture challenging scenarios in real-world built environments such as dynamic lighting, fast motions, and repetitive structural patterns with a low-cost solution. This dataset is primarily applied in construction progress monitoring, robotic autonomous navigation and computer vision fields, and is committed to promoting the practical development of visual SLAM and floor-plan-referenced localization technologies in complex industrial environments.
- 1Hilti-Trimble-Oxford Dataset: 360 Visual-Inertial Benchmark with Floor Plan Priors for SLAM and Localization喜利得股份公司·企业研究与技术部; 牛津大学·动态机器人系统组; 苏黎世联邦理工学院·机器人系统实验室; Vision & Robotics 有限公司; 天宝公司 · 2026年



