Dataset · MINT-SJTU
Real-world robot interaction video
MINT-SJTU/RW-RL-Dataset
Large video dataset of robot interactions with the real environment for reinforcement learning.
What's inside
The dataset contains between 100,000 and 1 million video modality records with English descriptions. The data was collected during real robots' interactions with the environment and released jointly by Boden Intelligence, Junpu Innovation Center, and the MINT laboratory of Shanghai Jiao Tong University.
What you can do with it
- Training reinforcement learning (RL) algorithms for real-world robotics
- Developing robot control systems based on visual data
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 417K
- Likes
- 19
Models trained on it
- Elvinky/pi05-piperx-8h-seed1000-fp32Robotics 26
- Elvinky/pi05-so101-stationery-v12-human1to1-fp32-step8553Robotics 18
- Elvinky/pi05-so101-fold-clothes-v11-human1to1-fp32-step8357Robotics 17
- Elvinky/pi05-piperx-2h-seed1000-fp32Robotics 16
- Elvinky/pi05-piperx-3h-intervention-1to1-mask-8884s-seed1000-fp32Robotics 16
- Elvinky/pi05-piperx-4h-seed1000-fp32Robotics 15
- Elvinky/pi05-piperx-7h-intervention-1to1-mask-9058s-seed1000-fp32Robotics 15
- Elvinky/pi05-piperx-1h-intervention-1to1-mask-8066s-seed1000-fp32Robotics 10
How to load it
from datasets import load_dataset
ds = load_dataset("MINT-SJTU/RW-RL-Dataset")