stanford_robocook_converted_externally_to_rlds

  • Description:

Franka preparing dumplings with various tools

Split Examples
'train' 2,460
  • Feature structure:
FeaturesDict({
    'episode_metadata': FeaturesDict({
        'extrinsics_1': Tensor(shape=(4, 4), dtype=float32, description=Camera 1 Extrinsic Matrix.),
        'extrinsics_2': Tensor(shape=(4, 4), dtype=float32, description=Camera 2 Extrinsic Matrix.),
        'extrinsics_3': Tensor(shape=(4, 4), dtype=float32, description=Camera 3 Extrinsic Matrix.),
        'extrinsics_4': Tensor(shape=(4, 4), dtype=float32, description=Camera 4 Extrinsic Matrix.),
        'file_path': Text(shape=(), dtype=string),
    }),
    'steps': Dataset({
        'action': Tensor(shape=(7,), dtype=float32, description=Robot action, consists of [3x robot end-effector velocities, 3x robot end-effector angular velocities, 1x gripper velocity].),
        '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': Tensor(shape=(256, 256), dtype=float32, description=Camera 1 Depth observation.),
            'depth_2': Tensor(shape=(256, 256), dtype=float32, description=Camera 2 Depth observation.),
            'depth_3': Tensor(shape=(256, 256), dtype=float32, description=Camera 3 Depth observation.),
            'depth_4': Tensor(shape=(256, 256), dtype=float32, description=Camera 4 Depth observation.),
            'image_1': Image(shape=(256, 256, 3), dtype=uint8, description=Camera 1 RGB observation.),
            'image_2': Image(shape=(256, 256, 3), dtype=uint8, description=Camera 2 RGB observation.),
            'image_3': Image(shape=(256, 256, 3), dtype=uint8, description=Camera 3 RGB observation.),
            'image_4': Image(shape=(256, 256, 3), dtype=uint8, description=Camera 4 RGB observation.),
            'state': Tensor(shape=(7,), dtype=float32, description=Robot state, consists of [3x robot end-effector position, 3x robot end-effector euler angles, 1x gripper position].),
        }),
        '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/extrinsics_1 Tensor (4, 4) float32 Camera 1 Extrinsic Matrix.
episode_metadata/extrinsics_2 Tensor (4, 4) float32 Camera 2 Extrinsic Matrix.
episode_metadata/extrinsics_3 Tensor (4, 4) float32 Camera 3 Extrinsic Matrix.
episode_metadata/extrinsics_4 Tensor (4, 4) float32 Camera 4 Extrinsic Matrix.
episode_metadata/file_path Text string Path to the original data file.
steps Dataset
steps/action Tensor (7,) float32 Robot action, consists of [3x robot end-effector velocities, 3x robot end-effector angular velocities, 1x gripper velocity].
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 Tensor (256, 256) float32 Camera 1 Depth observation.
steps/observation/depth_2 Tensor (256, 256) float32 Camera 2 Depth observation.
steps/observation/depth_3 Tensor (256, 256) float32 Camera 3 Depth observation.
steps/observation/depth_4 Tensor (256, 256) float32 Camera 4 Depth observation.
steps/observation/image_1 Image (256, 256, 3) uint8 Camera 1 RGB observation.
steps/observation/image_2 Image (256, 256, 3) uint8 Camera 2 RGB observation.
steps/observation/image_3 Image (256, 256, 3) uint8 Camera 3 RGB observation.
steps/observation/image_4 Image (256, 256, 3) uint8 Camera 4 RGB observation.
steps/observation/state Tensor (7,) float32 Robot state, consists of [3x robot end-effector position, 3x robot end-effector euler angles, 1x gripper position].
steps/reward Scalar float32 Reward if provided, 1 on final step for demos.
  • Citation:
@article{shi2023robocook,
  title={RoboCook: Long-Horizon Elasto-Plastic Object Manipulation with Diverse Tools},
  author={Shi, Haochen and Xu, Huazhe and Clarke, Samuel and Li, Yunzhu and Wu, Jiajun},
  journal={arXiv preprint arXiv:2306.14447},
  year={2023}
}