HHI-Assist
收藏资源简介:
HHI-Assist数据集由EPFL的VITA实验室创建,包含908段人类在辅助任务中的运动捕捉视频,涉及从椅子坐起、从床躺下坐起、从床躺下站起以及无约束动作等四种辅助任务。数据集由21个关节和20个链接组成的骨架模型采集,每秒120帧,数据以BVH格式存储。该数据集旨在帮助机器人预测人类运动,为机器人控制策略提供数据驱动的方法,并作为物理人机交互的性能基准。数据集和代码已公开。
The HHI-Assist dataset was created by the VITA Lab at EPFL. It contains 908 motion capture videos of human movements during assistive tasks, covering four types of assistive tasks: rising from a chair, sitting up from lying on a bed, standing up from lying on a bed, and unconstrained movements. The dataset is captured using a skeleton model consisting of 21 joints and 20 links, with a frame rate of 120 frames per second, and the data is stored in BVH format. This dataset aims to aid robots in predicting human motions, provide data-driven approaches for robot control strategies, and serve as a performance benchmark for physical human-robot interaction. The dataset and code have been made publicly available.

- 1通过VITA实验室,EPFL,瑞士洛桑 · 2025年



