the-future-devapache-2.0
PushT Dataset
A dataset of human demonstrations for the PushT environment, where an agent pushes a T-shaped object onto a matching target surface with randomized initial positions and orientations.
Downloads236
Episodes101
Why This Matters for Physical AI
This dataset enables training of imitation learning and reinforcement learning policies for object manipulation tasks with visual observations and 2D action control, contributing to the development of robotic pushing and placement skills.
Technical Profile
- Modalities
- rgbproprioception
- Robot Embodiments
- 2d pointer
- Action Space
- end_effector_delta
- Environment
- simulation
- Task Types
- manipulationpush_and_place
- Episodes
- 101
- Data Format
- parquet
- Annotation Types
- reward_labelsaction_labels
- License
- apache-2.0
Access
Need custom rgb data?
Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.
Request a Sample Pack