nyu_franka_play_dataset_converted_externally_to_rlds

  • 説明

おもちゃのキッチンと対話するフランカ

スプリット
'train' 365
'val' 91
  • 機能の構造:
FeaturesDict({
    '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),
    }),
})
  • 機能ドキュメント:
特徴クラスDタイプ説明
特徴辞書
エピソード_メタデータ特徴辞書
エピソードメタデータ/ファイルパス文章元のデータ ファイルへのパス。
ステップデータセット
ステップ/アクションテンソル(15,) float32ロボット アクションは、[7x 関節速度、3x EE デルタ xyz、3x EE デルタ rpy、1x グリッパー位置、1x 終了エピソード] で構成されます。
歩数/割引スカラーfloat32割引が指定されている場合、デフォルトは 1 です。
ステップ/is_firstテンソルブール
ステップ/is_lastテンソルブール
ステップ/is_terminalテンソルブール
ステップ/言語_埋め込みテンソル(512,) float32コナ言語の埋め込み。 https://tfhub.dev/google/universal-sentence-encoder-large/5を参照してください。
ステップ/言語説明文章言語指導。
ステップ/観察特徴辞書
歩数/観察/深度テンソル(128、128、1) int32右カメラの深度観察。
ステップ/観察/深さ_追加_ビューテンソル(128、128、1) int32左カメラの深度観察。
手順・観察・イメージ画像(128、128、3) uint8右カメラRGB観察。
ステップ/観察/画像追加_ビュー画像(128、128、3) uint8左カメラRGB観察。
ステップ/観察/状態テンソル(13,) float32ロボットの状態は、[7x ロボット関節角度、3x EE xyz、3x EE rpy] で構成されます。
歩数/報酬スカラーfloat32提供されている場合は報酬、デモの最終ステップで 1。
  • 引用
@article{cui2022play,
  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}
}