asu_table_top_converted_externally_to_rlds

  • Description:

UR5 performing table-top pick/place/rotate tasks

Split Examples
'train' 110
  • Feature structure:
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.),
    }),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
episode_metadata FeaturesDict
episode_metadata/file_path Text string Path to the original data file.
steps Dataset
steps/action Tensor (7,) float32 Robot action, consists of [7x joint velocities, 2x gripper velocities, 1x terminate episode].
steps/action_delta Tensor (7,) float32 Robot delta action, consists of [7x joint velocities, 2x gripper velocities, 1x terminate episode].
steps/action_inst Text string Action to be performed.
steps/discount Scalar float32 Discount if provided, default to 1.
steps/goal_object Text string Object to be manipulated with.
steps/ground_truth_states FeaturesDict
steps/ground_truth_states/EE Tensor (6,) float32 xyzrpy
steps/ground_truth_states/bottle Tensor (6,) float32 xyzrpy
steps/ground_truth_states/bread Tensor (6,) float32 xyzrpy
steps/ground_truth_states/coke Tensor (6,) float32 xyzrpy
steps/ground_truth_states/cube Tensor (6,) float32 xyzrpy
steps/ground_truth_states/milk Tensor (6,) float32 xyzrpy
steps/ground_truth_states/pepsi Tensor (6,) float32 xyzrpy
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/image Image (224, 224, 3) uint8 Main camera RGB observation.
steps/observation/state Tensor (7,) float32 Robot state, consists of [6x robot joint angles, 1x gripper position].
steps/observation/state_vel Tensor (7,) float32 Robot joint velocity, consists of [6x robot joint angles, 1x gripper position].
steps/reward Scalar float32 Reward if provided, 1 on final step for demos.
  • Citation:
@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}
}