HAPS 2.0
收藏资源简介:
HAPS 2.0数据集是由华盛顿大学等机构创建的,包含486个SMPL运动序列,覆盖了26种区域类型,如客厅、健身房、人行道等。数据集通过精细的标注流程,确保了运动描述与3D运动序列的准确对齐,适用于研究社会意识导航系统中的动态多人类交互。该数据集支持离散和连续导航环境,为机器人导航研究提供了丰富的多人类动态和部分可观测性挑战。
The HAPS 2.0 dataset was created by the University of Washington and other research institutions. It contains 486 SMPL motion sequences covering 26 types of regional environments, including living rooms, gyms, sidewalks, and more. Through a rigorous annotation pipeline, the dataset ensures accurate alignment between motion descriptions and 3D motion sequences, making it suitable for researching dynamic multi-human interactions in socially-aware navigation systems. This dataset supports both discrete and continuous navigation environments, offering rich multi-human dynamics and partially observable challenges for robotic navigation research.




