berkeley_gnm_sac_son

  • Description:

office navigation

Split Examples
'train' 2,955
  • Feature structure:
FeaturesDict({
    'episode_metadata': FeaturesDict({
        'file_path': Text(shape=(), dtype=string),
    }),
    'steps': Dataset({
        'action': Tensor(shape=(2,), dtype=float64),
        'action_angle': Tensor(shape=(3,), dtype=float64),
        'discount': Scalar(shape=(), dtype=float64),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'language_embedding': Tensor(shape=(512,), dtype=float32),
        'language_instruction': Text(shape=(), dtype=string),
        'observation': FeaturesDict({
            'image': Image(shape=(120, 160, 3), dtype=uint8),
            'position': Tensor(shape=(2,), dtype=float64),
            'state': Tensor(shape=(3,), dtype=float64),
            'yaw': Tensor(shape=(1,), dtype=float64),
        }),
        'reward': Scalar(shape=(), dtype=float64),
    }),
})
  • 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 (2,) float64 Robot action, consists of 2x position
steps/action_angle Tensor (3,) float64 Robot action, consists of 2x position, 1x yaw
steps/discount Scalar float64 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/image Image (120, 160, 3) uint8 Main camera RGB observation.
steps/observation/position Tensor (2,) float64 Robot position
steps/observation/state Tensor (3,) float64 Robot state, consists of [2x position, 1x yaw]
steps/observation/yaw Tensor (1,) float64 Robot yaw
steps/reward Scalar float64 Reward if provided, 1 on final step for demos.
  • Citation:
@article{hirose2023sacson,
  title={SACSoN: Scalable Autonomous Data Collection for Social Navigation},
  author={Hirose, Noriaki and Shah, Dhruv and Sridhar, Ajay and Levine, Sergey},
  journal={arXiv preprint arXiv:2306.01874},
  year={2023}
}