Tabletop Tasks Dataset
收藏arXiv2025-09-30 收录
下载链接：
http://dexterous-finetuning.github.io
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含9项涉及软质可变形物体的桌面操作任务，旨在评估机器人手的灵巧操作能力。这些任务包括但不限于拿取杯子、倾倒杯子、打开抽屉等，任务中物体位置随机，物体形状多样，以此测试操作的泛化能力。该数据集在9项任务中收集了超过100小时的真实世界数据，其研究重点是软物体的灵巧操作。
This dataset includes 9 tabletop manipulation tasks involving soft deformable objects, aiming to evaluate the dexterous manipulation capabilities of robotic hands. These tasks cover, but are not limited to, grasping cups, pouring from cups, opening drawers and more. The initial positions of objects are randomized and the objects have diverse shapes, so as to test the generalization ability of manipulation operations. Over 100 hours of real-world data have been collected across the 9 tasks, and the core research focus of this dataset is on dexterous manipulation of soft objects.
提供机构：
Authors of the paper
