stanford_hydra_dataset_converted_externally_to_rlds

  • Description:

Franka solving long-horizon tasks

Split Examples
'train' 570
  • Feature structure:
FeaturesDict({
    'episode_metadata': FeaturesDict({
        'file_path': Text(shape=(), dtype=string),
    }),
    'steps': Dataset({
        'action': Tensor(shape=(7,), dtype=float32, description=Robot action, consists of [3x EEF positional delta, 3x EEF orientation delta in euler angle, 1x close gripper].),
        'discount': Scalar(shape=(), dtype=float32, description=Discount if provided, default to 1.),
        'is_dense': Scalar(shape=(), dtype=bool, description=True if state is a waypoint(010) or in dense mode(x111).),
        '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({
            'image': Image(shape=(240, 320, 3), dtype=uint8, description=Main camera RGB observation.),
            'state': Tensor(shape=(27,), dtype=float32, description=Robot state, consists of [3x EEF position,4x EEF orientation in quaternion,3x EEF orientation in euler angle,7x robot joint angles, 7x robot joint velocities,3x gripper state.),
            'wrist_image': Image(shape=(240, 320, 3), dtype=uint8, description=Wrist camera RGB observation.),
        }),
        '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 (7,) float32 Robot action, consists of [3x EEF positional delta, 3x EEF orientation delta in euler angle, 1x close gripper].
steps/discount Scalar float32 Discount if provided, default to 1.
steps/is_dense Scalar bool True if state is a waypoint(010) or in dense mode(x111).
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/image Image (240, 320, 3) uint8 Main camera RGB observation.
steps/observation/state Tensor (27,) float32 Robot state, consists of [3x EEF position,4x EEF orientation in quaternion,3x EEF orientation in euler angle,7x robot joint angles, 7x robot joint velocities,3x gripper state.
steps/observation/wrist_image Image (240, 320, 3) uint8 Wrist camera RGB observation.
steps/reward Scalar float32 Reward if provided, 1 on final step for demos.
  • Citation:
@article{belkhale2023hydra,
 title={HYDRA: Hybrid Robot Actions for Imitation Learning},
 author={Belkhale, Suneel and Cui, Yuchen and Sadigh, Dorsa},
 journal={arxiv},
 year={2023}
}