• Description:

Franka interacting with toy kitchens

Split Examples
'train' 365
'val' 91
  • Feature structure:
    'episode_metadata': FeaturesDict({
        'file_path': Text(shape=(), dtype=string),
    'steps': Dataset({
        'action': Tensor(shape=(15,), dtype=float32),
        'discount': Scalar(shape=(), dtype=float32),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'language_embedding': Tensor(shape=(512,), dtype=float32),
        'language_instruction': Text(shape=(), dtype=string),
        'observation': FeaturesDict({
            'depth': Tensor(shape=(128, 128, 1), dtype=int32),
            'depth_additional_view': Tensor(shape=(128, 128, 1), dtype=int32),
            'image': Image(shape=(128, 128, 3), dtype=uint8),
            'image_additional_view': Image(shape=(128, 128, 3), dtype=uint8),
            'state': Tensor(shape=(13,), dtype=float32),
        'reward': Scalar(shape=(), dtype=float32),
  • Feature documentation:
Feature Class Shape Dtype Description
episode_metadata FeaturesDict
episode_metadata/file_path Text string Path to the original data file.
steps Dataset
steps/action Tensor (15,) float32 Robot action, consists of [7x joint velocities, 3x EE delta xyz, 3x EE delta rpy, 1x gripper position, 1x terminate episode].
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 Tensor (128, 128, 1) int32 Right camera depth observation.
steps/observation/depth_additional_view Tensor (128, 128, 1) int32 Left camera depth observation.
steps/observation/image Image (128, 128, 3) uint8 Right camera RGB observation.
steps/observation/image_additional_view Image (128, 128, 3) uint8 Left camera RGB observation.
steps/observation/state Tensor (13,) float32 Robot state, consists of [7x robot joint angles, 3x EE xyz, 3x EE rpy.
steps/reward Scalar float32 Reward if provided, 1 on final step for demos.
  • Citation:
  title   = {From Play to Policy: Conditional Behavior Generation from Uncurated Robot Data},
  author  = {Cui, Zichen Jeff and Wang, Yibin and Shafiullah, Nur Muhammad Mahi and Pinto, Lerrel},
  journal = {arXiv preprint arXiv:2210.10047},
  year    = {2022}