asu_table_top_converted_externally_to_rlds

  • 説明

卓上でピック/配置/回転タスクを実行する UR5

スプリット
'train' 110
  • 機能の構造:
FeaturesDict({
   
'episode_metadata': FeaturesDict({
       
'file_path': Text(shape=(), dtype=string),
   
}),
   
'steps': Dataset({
       
'action': Tensor(shape=(7,), dtype=float32, description=Robot action, consists of [7x joint velocities, 2x gripper velocities, 1x terminate episode].),
       
'action_delta': Tensor(shape=(7,), dtype=float32, description=Robot delta action, consists of [7x joint velocities, 2x gripper velocities, 1x terminate episode].),
       
'action_inst': Text(shape=(), dtype=string),
       
'discount': Scalar(shape=(), dtype=float32, description=Discount if provided, default to 1.),
       
'goal_object': Text(shape=(), dtype=string),
       
'ground_truth_states': FeaturesDict({
           
'EE': Tensor(shape=(6,), dtype=float32, description=xyzrpy),
           
'bottle': Tensor(shape=(6,), dtype=float32, description=xyzrpy),
           
'bread': Tensor(shape=(6,), dtype=float32, description=xyzrpy),
           
'coke': Tensor(shape=(6,), dtype=float32, description=xyzrpy),
           
'cube': Tensor(shape=(6,), dtype=float32, description=xyzrpy),
           
'milk': Tensor(shape=(6,), dtype=float32, description=xyzrpy),
           
'pepsi': Tensor(shape=(6,), dtype=float32, description=xyzrpy),
       
}),
       
'is_first': bool,
       
'is_last': bool,
       
'is_terminal': bool,
       
'language_embedding': Tensor(shape=(512,), dtype=float32, description=Kona language embedding. See https://tfhub.dev/google/universal-sentence-encoder-large/5),
       
'language_instruction': Text(shape=(), dtype=string),
       
'observation': FeaturesDict({
           
'image': Image(shape=(224, 224, 3), dtype=uint8, description=Main camera RGB observation.),
           
'state': Tensor(shape=(7,), dtype=float32, description=Robot state, consists of [6x robot joint angles, 1x gripper position].),
           
'state_vel': Tensor(shape=(7,), dtype=float32, description=Robot joint velocity, consists of [6x robot joint angles, 1x gripper position].),
       
}),
       
'reward': Scalar(shape=(), dtype=float32, description=Reward if provided, 1 on final step for demos.),
   
}),
})
  • 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、[7x 関節速度、2x グリッパー速度、1x 終了エピソード] で構成されます。
ステップ/アクションデルタテンソル(7,) float32ロボット デルタ アクションは、[7x ジョイント速度、2x グリッパー速度、1x 終了エピソード] で構成されます。
ステップ/アクション_インスタント文章実行されるアクション。
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/ゴール_オブジェクト文章操作するオブジェクト。
ステップ/ground_truth_states特徴辞書
ステップ/ground_truth_states/EEテンソル(6,) float32 xyzrpy
ステップ/ground_truth_states/ボトルテンソル(6,) float32 xyzrpy
ステップ/ground_truth_states/breadテンソル(6,) float32 xyzrpy
ステップ/ground_truth_states/cokeテンソル(6,) float32 xyzrpy
ステップ/ground_truth_states/cubeテンソル(6,) float32 xyzrpy
ステップ/ground_truth_states/milkテンソル(6,) float32 xyzrpy
ステップ/ground_truth_states/ペプシテンソル(6,) float32 xyzrpy
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(224, 224, 3) uint8メインカメラRGB観察。
ステップ/観察/状態テンソル(7,) float32ロボットの状態は、[6x ロボット関節角度、1x グリッパー位置] で構成されます。
ステップ/観察/state_velテンソル(7,) float32ロボット ジョイントの速度は、[6x ロボット ジョイント角度、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{zhou2023modularity,
  title
={Modularity through Attention: Efficient Training and Transfer of Language-Conditioned Policies for Robot Manipulation},
  author
={Zhou, Yifan and Sonawani, Shubham and Phielipp, Mariano and Stepputtis, Simon and Amor, Heni},
  booktitle
={Conference on Robot Learning},
  pages
={1684--1695},
  year
={2023},
  organization
={PMLR}
}
@article{zhou2023learning,
  title
={Learning modular language-conditioned robot policies through attention},
  author
={Zhou, Yifan and Sonawani, Shubham and Phielipp, Mariano and Ben Amor, Heni and Stepputtis, Simon},
  journal
={Autonomous Robots},
  pages
={1--21},
  year
={2023},
  publisher
={Springer}
}