Dataset · MINT-SJTU
Robot manipulator error analysis video dataset
MINT-SJTU/RoboFAC-dataset
Over 10 thousand videos and 78 thousand QA pairs for training robots to detect and fix errors.
What's inside
A multilingual VQA dataset in English from the RoboFAC research paper. Includes over 10 thousand video recordings of robot manipulations in simulation and the real world, supplemented by 78,623 question-answer pairs for analyzing failure causes and their correction.
What you can do with it
- Training multimodal video-text-to-text models for robot failure diagnosis
- Creating self-correction action systems for robotic manipulators
- Tasks
- Video text → text
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 22K
- Likes
- 8
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("MINT-SJTU/RoboFAC-dataset")