TouchScale: 500 Hours of Human Vision and Touch for Visual-Tactile Learning
把人类双手的「手感」移植给机器人：500 小时真实操作数据
TouchScale is a large-scale human vision-touch dataset that enables scaling benefits in both perception and robotic manipulation. It uses a unified sensor setup and provides high-quality data for training robots to perform contact-intensive tasks with improved success rates.
Source: Chinese robotics — Hardware and sensing · Read original article ↗
English translation unavailable. Choose the original article to read it.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
Reported duration
500 hours
Reported duration
500 hours
Reported duration
500 hours
Reported duration
30 hours
taxel
880
Reported success rate
22.5%
Reported success rate
57.5%
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: Available Artifact link Open source S19
- paper: Available Artifact link Open source S19
What remains unknown
Not established in the collected evidence: Environment, Control, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
把人类双手的「手感」移植给机器人：500 小时真实操作数据
Open source S1
TouchScale 包含 500 小时人类双手视觉-触觉同步数据，全部由同一套穿戴式系统采集，同步记录头部 RGB-D、双腕 RGB 和双手全手触觉信号。
Open source S13
题目：TouchScale: 500 Hours of Human Vision and Touch for Visual-Tactile Learning 论文链接：https://arxiv.org/abs/2610.10288 项目主页：https://touch-scale.github.io/ 数据集地址：https://huggingface.co/datasets/2077AIDataFoundation/TouchScale 500 小时触觉数据，难的不只是「采得多」 对 TouchScale 来说，真正的难点不是把小时数简单堆到 500，而是在同一套传感器、同一套同步与采集流程下持续扩展。
Open source S19
此前公开的人类视觉-触觉数据集大多不足 30 小时；如果直接把多个小数据集拼在一起，不同硬件和采集流程会与数据规模一起变化，很难把规模本身的影响单独拿出来研究。
Open source S20
与已有公开的人类视觉-触觉数据相比，TouchScale 不仅将规模扩展到 500 小时，还始终保持统一的传感器和采集流程，并提供双手全手触觉采集，每只手包含 880 个 taxel。
Open source S23
无需人类动作标签，也不把人手动作映射到机器人，只用人类视触觉数据做 mid-training，机器人 4 个真机接触任务的平均成功率从 22.5% 升到 57.5%。
Open source S15
Source：Chinese robotics — Hardware and sensing · 163.com