Sterzhang
MotionLM motion question answering
A visual question-answering dataset combining MotionLM training and test exports with 8,202 train and 1,000 test examples, each containing a multiple-choice question about robot motion and 16 chronological RGB frames from robotic manipulation tasks.
Downloads66
Episodes9202
Why This Matters for Physical AI
This dataset enables training of multimodal models to understand and reason about robot motion sequences, bridging vision and language understanding for robotic task comprehension.
Technical Profile
- Modalities
- rgblanguage
- Environment
- lab
- Task Types
- motion-understandingvisual-question-answering
- Episodes
- 9202
- Data Format
- Parquet
- Annotation Types
- language_instructionsaction_labels
Access
Need custom rgb data?
Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.
Request a Sample Pack