tokyo_u_lsmo_converted_externally_to_rlds

  • 説明

ピック・プレイス・タスクの動作計画軌道

スプリット
'train' 50
  • 機能の構造:
FeaturesDict({
    'episode_metadata': FeaturesDict({
        'file_path': Text(shape=(), dtype=string),
    }),
    'steps': Dataset({
        'action': Tensor(shape=(7,), 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=(120, 120, 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(7、) float32ロボットアクションは、[3x エンドエフェクター位置、3x オイラー角、1x グリッパーアクション] 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(120、120、3) uint8メインカメラ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、[3x エンドエフェクター位置、3x オイラー角、6x 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。
@Article{Osa22,
  author  = {Takayuki Osa},
  journal = {The International Journal of Robotics Research},
  title   = {Motion Planning by Learning the Solution Manifold in Trajectory Optimization},
  year    = {2022},
  number  = {3},
  pages   = {291--311},
  volume  = {41},
}