cmu_play_fusion

  • 説明

このロボットは 3 つの複雑なシーンで動作します。つまり、トースターやフライパンなどの調理対象物がたくさんあるグリルです。選ぶ、開く、置く、閉じるという操作を行う必要があります。テーブルをセットし、皿、カップ、調理器具を移動しなければなりません。そして、シンク、食器洗い機、ハンドカップなどに食器を置く必要があります。

スプリット
'train' 576
  • 機能の構造:
FeaturesDict({
    'episode_metadata': FeaturesDict({
        'file_path': Text(shape=(), dtype=string),
    }),
    'steps': Dataset({
        'action': Tensor(shape=(9,), 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({
            'image': Image(shape=(128, 128, 3), dtype=uint8),
            'state': Tensor(shape=(8,), 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(9,) float32ロボット アクションは、[7x デルタ eef (pos + quat)、1x グリッパー開閉 (バイナリ)、1x エピソード終了] 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
手順・観察・イメージ画像(128、128、3) uint8メインカメラRGB観察。
ステップ/観察/状態テンソル(8,) float32ロボットの状態は、[7x ロボット関節角度、1x グリッパー位置] 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
@inproceedings{chen2023playfusion,
  title={PlayFusion: Skill Acquisition via Diffusion from Language-Annotated Play},
  author={Chen, Lili and Bahl, Shikhar and Pathak, Deepak},
  booktitle={CoRL},
  year={2023}
}