GrandTour
收藏资源简介:
GrandTour是由苏黎世联邦理工学院等机构联合开发的多模态腿式机器人数据集,旨在解决复杂环境中自主机器人的状态估计与感知问题。该数据集包含49条任务序列,总长度超过10公里,数据量超过5小时,采集自ANYmal-D四足机器人搭载的Boxi多模态传感器套件,涵盖LiDAR、多视角RGB相机、深度相机、IMU及高精度RTK-GNSS地面真值。数据覆盖高山、森林、城市废墟等多样化场景,包含不同光照与天气条件。其核心价值在于为SLAM、多模态学习及传感器融合研究提供真实世界基准,支持腿式机器人导航算法的开发与验证。
GrandTour is a multimodal legged robotics dataset co-developed by ETH Zurich and other institutions, aiming to address the state estimation and perception challenges of autonomous robots in complex environments. This dataset contains 49 task sequences, with a total traversed distance of over 10 kilometers and a total data duration of more than 5 hours. It is collected from the Boxi multimodal sensor suite mounted on the ANYmal-D quadruped robot, covering LiDAR, multi-view RGB cameras, depth cameras, IMU, and high-precision RTK-GNSS ground truth. The data covers diverse scenarios including alpine areas, forests, and urban ruins, and includes various lighting and weather conditions. Its core value is to provide real-world benchmarks for SLAM, multimodal learning and sensor fusion research, supporting the development and validation of legged robot navigation algorithms.



