uiuc_d3field

  • Description:

Organizing office desk, utensils etc

Split Examples
'train' 192
  • Feature structure:
FeaturesDict({
    'episode_metadata': FeaturesDict({
        'file_path': Text(shape=(), dtype=string),
    }),
    'steps': Dataset({
        'action': Tensor(shape=(3,), dtype=float32, description=Robot displacement from last frame),
        'discount': Scalar(shape=(), dtype=float32, description=Discount if provided, default to 1.),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'language_embedding': Tensor(shape=(512,), dtype=float32, description=Kona language embedding. See https://tfhub.dev/google/universal-sentence-encoder-large/5),
        'language_instruction': Text(shape=(), dtype=string),
        'observation': FeaturesDict({
            'depth_1': Image(shape=(360, 640, 1), dtype=uint16, description=camera 1 depth observation.),
            'depth_2': Image(shape=(360, 640, 1), dtype=uint16, description=camera 2 depth observation.),
            'depth_3': Image(shape=(360, 640, 1), dtype=uint16, description=camera 3 depth observation.),
            'depth_4': Image(shape=(360, 640, 1), dtype=uint16, description=camera 4 depth observation.),
            'image_1': Image(shape=(360, 640, 3), dtype=uint8, description=camera 1 RGB observation.),
            'image_2': Image(shape=(360, 640, 3), dtype=uint8, description=camera 2 RGB observation.),
            'image_3': Image(shape=(360, 640, 3), dtype=uint8, description=camera 3 RGB observation.),
            'image_4': Image(shape=(360, 640, 3), dtype=uint8, description=camera 4 RGB observation.),
            'state': Tensor(shape=(4, 4), dtype=float32, description=Robot end-effector state),
        }),
        'reward': Scalar(shape=(), dtype=float32, description=Reward if provided, 1 on final step for demos.),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_metadata FeaturesDict
episode_metadata/file_path Text string Path to the original data file.
steps Dataset
steps/action Tensor (3,) float32 Robot displacement from last frame
steps/discount Scalar float32 Discount if provided, default to 1.
steps/is_first Tensor bool
steps/is_last Tensor bool
steps/is_terminal Tensor bool
steps/language_embedding Tensor (512,) float32 Kona language embedding. See https://tfhub.dev/google/universal-sentence-encoder-large/5
steps/language_instruction Text string Language Instruction.
steps/observation FeaturesDict
steps/observation/depth_1 Image (360, 640, 1) uint16 camera 1 depth observation.
steps/observation/depth_2 Image (360, 640, 1) uint16 camera 2 depth observation.
steps/observation/depth_3 Image (360, 640, 1) uint16 camera 3 depth observation.
steps/observation/depth_4 Image (360, 640, 1) uint16 camera 4 depth observation.
steps/observation/image_1 Image (360, 640, 3) uint8 camera 1 RGB observation.
steps/observation/image_2 Image (360, 640, 3) uint8 camera 2 RGB observation.
steps/observation/image_3 Image (360, 640, 3) uint8 camera 3 RGB observation.
steps/observation/image_4 Image (360, 640, 3) uint8 camera 4 RGB observation.
steps/observation/state Tensor (4, 4) float32 Robot end-effector state
steps/reward Scalar float32 Reward if provided, 1 on final step for demos. 
@article{wang2023d3field,
  title={D^3Field: Dynamic 3D Descriptor Fields for Generalizable Robotic Manipulation},
  author={Wang, Yixuan and Li, Zhuoran and Zhang, Mingtong and Driggs-Campbell, Katherine and Wu, Jiajun and Fei-Fei, Li and Li, Yunzhu},
  journal={arXiv preprint arXiv:},
  year={2023},
}