| natural_language_instruction | image (256, 320, 3) |
|---|---|
| place green can upright | |
| pick apple from white bowl | |
| move blue chip bag near water bottle |
tfds.core.DatasetInfo(
name='fractal20220817_data',
full_name='fractal20220817_data/0.1.0',
description="""
Table-top manipulation with 17 objects
""",
homepage='https://ai.googleblog.com/2022/12/rt-1-robotics-transformer-for-real.html',
data_path='gs://gresearch/robotics/fractal20220817_data/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=111.07 GiB,
features=FeaturesDict({
'aspects': FeaturesDict({
'already_success': bool,
'feasible': bool,
'has_aspects': bool,
'success': bool,
'undesirable': bool,
}),
'attributes': FeaturesDict({
'collection_mode': int64,
'collection_mode_name': string,
'data_type': int64,
'data_type_name': string,
'env': int64,
'env_name': string,
'location': int64,
'location_name': string,
'objects_family': int64,
'objects_family_name': string,
'task_family': int64,
'task_family_name': string,
}),
'steps': Dataset({
'action': FeaturesDict({
'base_displacement_vector': Tensor(shape=(2,), dtype=float32),
'base_displacement_vertical_rotation': Tensor(shape=(1,), dtype=float32),
'gripper_closedness_action': Tensor(shape=(1,), dtype=float32),
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': Tensor(shape=(3,), dtype=int32),
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'base_pose_tool_reached': Tensor(shape=(7,), dtype=float32),
'gripper_closed': Tensor(shape=(1,), dtype=float32),
'gripper_closedness_commanded': Tensor(shape=(1,), dtype=float32),
'height_to_bottom': Tensor(shape=(1,), dtype=float32),
'image': Image(shape=(256, 320, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'orientation_box': Tensor(shape=(2, 3), dtype=float32),
'orientation_start': Tensor(shape=(4,), dtype=float32),
'robot_orientation_positions_box': Tensor(shape=(3, 3), dtype=float32),
'rotation_delta_to_go': Tensor(shape=(3,), dtype=float32),
'src_rotation': Tensor(shape=(4,), dtype=float32),
'vector_to_go': Tensor(shape=(3,), dtype=float32),
'workspace_bounds': Tensor(shape=(3, 3), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{brohan2022rt,
title={Rt-1: Robotics transformer for real-world control at scale},
author={Brohan, Anthony and Brown, Noah and Carbajal, Justice and Chebotar, Yevgen and Dabis, Joseph and Finn, Chelsea and Gopalakrishnan, Keerthana and Hausman, Karol and Herzog, Alex and Hsu, Jasmine and others},
journal={arXiv preprint arXiv:2212.06817},
year={2022}
}""",
)
| natural_language_instruction | image (512, 640, 3) |
|---|---|
| pick anything | |
| pick anything | |
| pick anything |
tfds.core.DatasetInfo(
name='kuka',
full_name='kuka/0.1.0',
description="""
Bin picking and rearrangement tasks
""",
homepage='https://arxiv.org/abs/1806.10293',
data_path='gs://gresearch/robotics/kuka/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=778.02 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'base_displacement_vector': Tensor(shape=(2,), dtype=float32),
'base_displacement_vertical_rotation': Tensor(shape=(1,), dtype=float32),
'gripper_closedness_action': Tensor(shape=(1,), dtype=float32),
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': Tensor(shape=(3,), dtype=int32),
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'clip_function_input/base_pose_tool_reached': Tensor(shape=(7,), dtype=float32),
'clip_function_input/workspace_bounds': Tensor(shape=(3, 3), dtype=float32),
'gripper_closed': Tensor(shape=(1,), dtype=float32),
'height_to_bottom': Tensor(shape=(1,), dtype=float32),
'image': Image(shape=(512, 640, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'task_id': Tensor(shape=(1,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
'success': bool,
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{kalashnikov2018qt,
title={Qt-opt: Scalable deep reinforcement learning for vision-based robotic manipulation},
author={Kalashnikov, Dmitry and Irpan, Alex and Pastor, Peter and Ibarz, Julian and Herzog, Alexander and Jang, Eric and Quillen, Deirdre and Holly, Ethan and Kalakrishnan, Mrinal and Vanhoucke, Vincent and others},
journal={arXiv preprint arXiv:1806.10293},
year={2018}
}""",
)
| natural_language_instruction | image (480, 640, 3) |
|---|---|
| PICK UP THE SPOON AND PUT NEAR THE VESSEL | |
| put fork from basket to tray | |
| Place the red vegetable in the silver pot. |
tfds.core.DatasetInfo(
name='bridge',
full_name='bridge/0.1.0',
description="""
WidowX interacting with toy kitchens
""",
homepage='https://rail-berkeley.github.io/bridgedata/',
data_path='gs://gresearch/robotics/bridge/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=387.49 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'open_gripper': bool,
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'image': Image(shape=(480, 640, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'state': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@inproceedings{walke2023bridgedata,
title={BridgeData V2: A Dataset for Robot Learning at Scale},
author={Walke, Homer and Black, Kevin and Lee, Abraham and Kim, Moo Jin and Du, Max and Zheng, Chongyi and Zhao, Tony and Hansen-Estruch, Philippe and Vuong, Quan and He, Andre and Myers, Vivek and Fang, Kuan and Finn, Chelsea and Levine, Sergey},
booktitle={Conference on Robot Learning (CoRL)},
year={2023}
}""",
)
| structured_language_instruction | natural_language_instruction | rgb_gripper (84, 84, 3) | rgb_static (150, 200, 3) |
|---|---|---|---|
| place_yellow_left_cabinet | put the yellow object inside the left cabinet | ||
| place_pink_box | move to the box and place the pink object | ||
| rotate_yellow_block_left | grasp the yellow block and turn it left |
tfds.core.DatasetInfo(
name='taco_play',
full_name='taco_play/0.1.0',
description="""
Franka arm interacting with kitchen
""",
homepage='https://www.kaggle.com/datasets/oiermees/taco-robot',
data_path='gs://gresearch/robotics/taco_play/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=47.77 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'actions': Tensor(shape=(7,), dtype=float32),
'rel_actions_gripper': Tensor(shape=(7,), dtype=float32),
'rel_actions_world': Tensor(shape=(7,), dtype=float32),
'terminate_episode': float32,
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'depth_gripper': Tensor(shape=(84, 84), dtype=float32),
'depth_static': Tensor(shape=(150, 200), dtype=float32),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'rgb_gripper': Image(shape=(84, 84, 3), dtype=uint8),
'rgb_static': Image(shape=(150, 200, 3), dtype=uint8),
'robot_obs': Tensor(shape=(15,), dtype=float32),
'structured_language_instruction': string,
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@inproceedings{rosete2022tacorl,
author = {Erick Rosete-Beas and Oier Mees and Gabriel Kalweit and Joschka Boedecker and Wolfram Burgard},
title = {Latent Plans for Task Agnostic Offline Reinforcement Learning},
journal = {Proceedings of the 6th Conference on Robot Learning (CoRL)},
year = {2022}
}
@inproceedings{mees23hulc2,
title={Grounding Language with Visual Affordances over Unstructured Data},
author={Oier Mees and Jessica Borja-Diaz and Wolfram Burgard},
booktitle = {Proceedings of the IEEE International Conference on Robotics and Automation (ICRA)},
year={2023},
address = {London, UK}
}""",
)
| natural_language_instruction | image (224, 224, 3) | image_wrist (224, 224, 3) |
|---|---|---|
| pick up the milk dairy | ||
| place the black bowl in the oven | ||
| pick up the green cup |
tfds.core.DatasetInfo(
name='jaco_play',
full_name='jaco_play/0.1.0',
description="""
Jaco 2 pick place on table top
""",
homepage='https://github.com/clvrai/clvr_jaco_play_dataset',
data_path='gs://gresearch/robotics/jaco_play/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=9.24 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'gripper_closedness_action': Tensor(shape=(1,), dtype=float32),
'terminate_episode': Tensor(shape=(3,), dtype=int32),
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'end_effector_cartesian_pos': Tensor(shape=(7,), dtype=float32),
'end_effector_cartesian_velocity': Tensor(shape=(6,), dtype=float32),
'image': Image(shape=(224, 224, 3), dtype=uint8),
'image_wrist': Image(shape=(224, 224, 3), dtype=uint8),
'joint_pos': Tensor(shape=(8,), dtype=float32),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@software{dass2023jacoplay,
author = {Dass, Shivin and Yapeter, Jullian and Zhang, Jesse and Zhang, Jiahui
and Pertsch, Karl and Nikolaidis, Stefanos and Lim, Joseph J.},
title = {CLVR Jaco Play Dataset},
url = {https://github.com/clvrai/clvr_jaco_play_dataset},
version = {1.0.0},
year = {2023}
}""",
)
| natural_language_instruction | wrist45_image (128, 128, 3) | image (128, 128, 3) | wrist225_image (128, 128, 3) | top_image (128, 128, 3) |
|---|---|---|---|---|
| route cable | ||||
| route cable | ||||
| route cable |
tfds.core.DatasetInfo(
name='berkeley_cable_routing',
full_name='berkeley_cable_routing/0.1.0',
description="""
Routing cable into clamps on table top
""",
homepage='https://sites.google.com/view/cablerouting/home',
data_path='gs://gresearch/robotics/berkeley_cable_routing/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=4.67 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'image': Image(shape=(128, 128, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'robot_state': Tensor(shape=(7,), dtype=float32),
'top_image': Image(shape=(128, 128, 3), dtype=uint8),
'wrist225_image': Image(shape=(128, 128, 3), dtype=uint8),
'wrist45_image': Image(shape=(128, 128, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@article{luo2023multistage,
author = {Jianlan Luo and Charles Xu and Xinyang Geng and Gilbert Feng and Kuan Fang and Liam Tan and Stefan Schaal and Sergey Levine},
title = {Multi-Stage Cable Routing through Hierarchical Imitation Learning},
journal = {arXiv pre-print},
year = {2023},
url = {https://arxiv.org/abs/2307.08927},
}""",
)
| natural_language_instruction | front_rgb (480, 640, 3) |
|---|---|
| object search | |
| object search | |
| layout laundry |
tfds.core.DatasetInfo(
name='roboturk',
full_name='roboturk/0.1.0',
description="""
Cloth folding, bowl stacking
""",
homepage='https://roboturk.stanford.edu/dataset_real.html',
data_path='gs://gresearch/robotics/roboturk/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=45.39 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'gripper_closedness_action': Tensor(shape=(1,), dtype=float32),
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'front_rgb': Image(shape=(480, 640, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@inproceedings{mandlekar2019scaling,
title={Scaling robot supervision to hundreds of hours with roboturk: Robotic manipulation dataset through human reasoning and dexterity},
author={Mandlekar, Ajay and Booher, Jonathan and Spero, Max and Tung, Albert and Gupta, Anchit and Zhu, Yuke and Garg, Animesh and Savarese, Silvio and Fei-Fei, Li},
booktitle={2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
pages={1048--1055},
year={2019},
organization={IEEE}
}""",
)
| natural_language_instruction | image (720, 960, 3) |
|---|---|
| open door | |
| open door | |
| open door |
tfds.core.DatasetInfo(
name='nyu_door_opening_surprising_effectiveness',
full_name='nyu_door_opening_surprising_effectiveness/0.1.0',
description="""
Hello robot opening cabinets, microwaves etc
""",
homepage='https://jyopari.github.io/VINN/',
data_path='gs://gresearch/robotics/nyu_door_opening_surprising_effectiveness/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=7.12 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'gripper_closedness_action': Tensor(shape=(1,), dtype=float32),
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'image': Image(shape=(720, 960, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@misc{pari2021surprising,
title={The Surprising Effectiveness of Representation Learning for Visual Imitation},
author={Jyothish Pari and Nur Muhammad Shafiullah and Sridhar Pandian Arunachalam and Lerrel Pinto},
year={2021},
eprint={2112.01511},
archivePrefix={arXiv},
primaryClass={cs.RO}
}""",
)
| natural_language_instruction | eye_in_hand_rgb (224, 224, 3) | agentview_rgb (224, 224, 3) |
|---|---|---|
| arrange plate and fork | ||
| arrange plate and fork | ||
| make coffee |
tfds.core.DatasetInfo(
name='viola',
full_name='viola/0.1.0',
description="""
Franka robot interacting with stylized kitchen tasks
""",
homepage='https://ut-austin-rpl.github.io/VIOLA/',
data_path='gs://gresearch/robotics/viola/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=10.40 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'gripper_closedness_action': float32,
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'agentview_rgb': Image(shape=(224, 224, 3), dtype=uint8),
'ee_states': Tensor(shape=(16,), dtype=float32),
'eye_in_hand_rgb': Image(shape=(224, 224, 3), dtype=uint8),
'gripper_states': Tensor(shape=(1,), dtype=float32),
'joint_states': Tensor(shape=(7,), dtype=float32),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@article{zhu2022viola,
title={VIOLA: Imitation Learning for Vision-Based Manipulation with Object Proposal Priors},
author={Zhu, Yifeng and Joshi, Abhishek and Stone, Peter and Zhu, Yuke},
journal={6th Annual Conference on Robot Learning (CoRL)},
year={2022}
}""",
)
| natural_language_instruction | image (480, 640, 3) | hand_image (480, 640, 3) |
|---|---|---|
| pick up the blue cup and put it into the brown cup. | ||
| put the ranch bottle into the pot | ||
| put the ranch bottle into the pot |
tfds.core.DatasetInfo(
name='berkeley_autolab_ur5',
full_name='berkeley_autolab_ur5/0.1.0',
description="""
UR5 performing cloth manipulation, pick place etc tasks
""",
homepage='https://sites.google.com/view/berkeley-ur5/home',
data_path='gs://gresearch/robotics/berkeley_autolab_ur5/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=76.39 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'gripper_closedness_action': float32,
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'hand_image': Image(shape=(480, 640, 3), dtype=uint8),
'image': Image(shape=(480, 640, 3), dtype=uint8),
'image_with_depth': Image(shape=(480, 640, 1), dtype=float32),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'robot_state': Tensor(shape=(15,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@misc{BerkeleyUR5Website,
title = {Berkeley {UR5} Demonstration Dataset},
author = {Lawrence Yunliang Chen and Simeon Adebola and Ken Goldberg},
howpublished = {https://sites.google.com/view/berkeley-ur5/home},
}""",
)
| natural_language_instruction | image (480, 640, 3) |
|---|---|
| pour | |
| pour | |
| pour |
tfds.core.DatasetInfo(
name='toto',
full_name='toto/0.1.0',
description="""
Franka scooping and pouring tasks
""",
homepage='https://toto-benchmark.org/',
data_path='gs://gresearch/robotics/toto/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=127.66 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'open_gripper': bool,
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'image': Image(shape=(480, 640, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'state': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@inproceedings{zhou2023train,
author={Zhou, Gaoyue and Dean, Victoria and Srirama, Mohan Kumar and Rajeswaran, Aravind and Pari, Jyothish and Hatch, Kyle and Jain, Aryan and Yu, Tianhe and Abbeel, Pieter and Pinto, Lerrel and Finn, Chelsea and Gupta, Abhinav},
booktitle={2023 IEEE International Conference on Robotics and Automation (ICRA)},
title={Train Offline, Test Online: A Real Robot Learning Benchmark},
year={2023},
}""",
)
| rgb (360, 640, 3) |
|---|
tfds.core.DatasetInfo(
name='language_table',
full_name='language_table/0.0.1',
description="""
""",
homepage='https://www.tensorflow.org/datasets/catalog/language_table',
data_path='gs://gresearch/robotics/language_table/0.0.1',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=399.23 GiB,
features=FeaturesDict({
'episode_id': string,
'steps': Dataset({
'action': Tensor(shape=(2,), dtype=float32),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'effector_target_translation': Tensor(shape=(2,), dtype=float32),
'effector_translation': Tensor(shape=(2,), dtype=float32),
'instruction': Tensor(shape=(512,), dtype=int32),
'rgb': Image(shape=(360, 640, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""""",
)
| natural_language_instruction | wrist_image (240, 320, 3) | image (240, 320, 3) |
|---|---|---|
| The task requires pushing a T-shaped block (gray) to a fixed target (green) with a circular end-effector (blue). Both observation and control frequencies are 10Hz. | ||
| The task requires pushing a T-shaped block (gray) to a fixed target (green) with a circular end-effector (blue). Both observation and control frequencies are 10Hz. | ||
| The task requires pushing a T-shaped block (gray) to a fixed target (green) with a circular end-effector (blue). Both observation and control frequencies are 10Hz. |
tfds.core.DatasetInfo(
name='columbia_cairlab_pusht_real',
full_name='columbia_cairlab_pusht_real/0.1.0',
description="""
UR5 planar pushing tasks
""",
homepage='https://github.com/columbia-ai-robotics/diffusion_policy',
data_path='gs://gresearch/robotics/columbia_cairlab_pusht_real/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=2.80 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'gripper_closedness_action': float32,
'rotation_delta': Tensor(shape=(3,), dtype=float32),
'terminate_episode': float32,
'world_vector': Tensor(shape=(3,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'image': Image(shape=(240, 320, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'robot_state': Tensor(shape=(2,), dtype=float32),
'wrist_image': Image(shape=(240, 320, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'test': ,
'train': ,
},
citation="""@inproceedings{chi2023diffusionpolicy,
title={Diffusion Policy: Visuomotor Policy Learning via Action Diffusion},
author={Chi, Cheng and Feng, Siyuan and Du, Yilun and Xu, Zhenjia and Cousineau, Eric and Burchfiel, Benjamin and Song, Shuran},
booktitle={Proceedings of Robotics: Science and Systems (RSS)},
year={2023}
}""",
)
| image (128, 128, 3) |
|---|
tfds.core.DatasetInfo(
name='stanford_kuka_multimodal_dataset_converted_externally_to_rlds',
full_name='stanford_kuka_multimodal_dataset_converted_externally_to_rlds/0.1.0',
description="""
Kuka iiwa peg insertion with force feedback
""",
homepage='https://sites.google.com/view/visionandtouch',
data_path='gs://gresearch/robotics/stanford_kuka_multimodal_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=31.98 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
}),
'steps': Dataset({
'action': Tensor(shape=(4,), 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({
'contact': Tensor(shape=(50,), dtype=float32),
'depth_image': Tensor(shape=(128, 128, 1), dtype=float32),
'ee_forces_continuous': Tensor(shape=(50, 6), dtype=float32),
'ee_orientation': Tensor(shape=(4,), dtype=float32),
'ee_orientation_vel': Tensor(shape=(3,), dtype=float32),
'ee_position': Tensor(shape=(3,), dtype=float32),
'ee_vel': Tensor(shape=(3,), dtype=float32),
'ee_yaw': Tensor(shape=(4,), dtype=float32),
'ee_yaw_delta': Tensor(shape=(4,), dtype=float32),
'image': Image(shape=(128, 128, 3), dtype=uint8),
'joint_pos': Tensor(shape=(7,), dtype=float32),
'joint_vel': Tensor(shape=(7,), dtype=float32),
'optical_flow': Tensor(shape=(128, 128, 2), dtype=float32),
'state': Tensor(shape=(8,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{lee2019icra,
title={Making sense of vision and touch: Self-supervised learning of multimodal representations for contact-rich tasks},
author={Lee, Michelle A and Zhu, Yuke and Srinivasan, Krishnan and Shah, Parth and Savarese, Silvio and Fei-Fei, Li and Garg, Animesh and Bohg, Jeannette},
booktitle={2019 IEEE International Conference on Robotics and Automation (ICRA)},
year={2019},
url={https://arxiv.org/abs/1810.10191}
}""",
)
| image (84, 84, 3) |
|---|
tfds.core.DatasetInfo(
name='nyu_rot_dataset_converted_externally_to_rlds',
full_name='nyu_rot_dataset_converted_externally_to_rlds/0.1.0',
description="""
xArm short-horizon table-top tasks
""",
homepage='https://rot-robot.github.io/',
data_path='gs://gresearch/robotics/nyu_rot_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=5.33 MiB,
features=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=(84, 84, 3), dtype=uint8),
'state': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{haldar2023watch,
title={Watch and match: Supercharging imitation with regularized optimal transport},
author={Haldar, Siddhant and Mathur, Vaibhav and Yarats, Denis and Pinto, Lerrel},
booktitle={Conference on Robot Learning},
pages={32--43},
year={2023},
organization={PMLR}
}""",
)
| image (240, 320, 3) | wrist_image (240, 320, 3) |
|---|---|
tfds.core.DatasetInfo(
name='stanford_hydra_dataset_converted_externally_to_rlds',
full_name='stanford_hydra_dataset_converted_externally_to_rlds/0.1.0',
description="""
Franka solving long-horizon tasks
""",
homepage='https://sites.google.com/view/hydra-il-2023',
data_path='gs://gresearch/robotics/stanford_hydra_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=72.48 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(7,), dtype=float32),
'discount': Scalar(shape=(), dtype=float32),
'is_dense': Scalar(shape=(), dtype=bool),
'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=(240, 320, 3), dtype=uint8),
'state': Tensor(shape=(27,), dtype=float32),
'wrist_image': Image(shape=(240, 320, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{belkhale2023hydra,
title={HYDRA: Hybrid Robot Actions for Imitation Learning},
author={Belkhale, Suneel and Cui, Yuchen and Sadigh, Dorsa},
journal={arxiv},
year={2023}
}""",
)
| image (128, 128, 3) | wrist_image (128, 128, 3) |
|---|---|
tfds.core.DatasetInfo(
name='austin_buds_dataset_converted_externally_to_rlds',
full_name='austin_buds_dataset_converted_externally_to_rlds/0.1.0',
description="""
Franka stylized kitchen tasks
""",
homepage='https://ut-austin-rpl.github.io/rpl-BUDS/',
data_path='gs://gresearch/robotics/austin_buds_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=1.49 GiB,
features=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=(128, 128, 3), dtype=uint8),
'state': Tensor(shape=(24,), dtype=float32),
'wrist_image': Image(shape=(128, 128, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{zhu2022bottom,
title={Bottom-Up Skill Discovery From Unsegmented Demonstrations for Long-Horizon Robot Manipulation},
author={Zhu, Yifeng and Stone, Peter and Zhu, Yuke},
journal={IEEE Robotics and Automation Letters},
volume={7},
number={2},
pages={4126--4133},
year={2022},
publisher={IEEE}
}""",
)
| image_additional_view (128, 128, 3) | image (128, 128, 3) |
|---|---|
tfds.core.DatasetInfo(
name='nyu_franka_play_dataset_converted_externally_to_rlds',
full_name='nyu_franka_play_dataset_converted_externally_to_rlds/0.1.0',
description="""
Franka interacting with toy kitchens
""",
homepage='https://play-to-policy.github.io/',
data_path='gs://gresearch/robotics/nyu_franka_play_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=5.18 GiB,
features=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),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""@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}
}""",
)
| wrist_image (256, 256, 3) | image (256, 256, 3) |
|---|---|
tfds.core.DatasetInfo(
name='maniskill_dataset_converted_externally_to_rlds',
full_name='maniskill_dataset_converted_externally_to_rlds/0.1.0',
description="""
Simulated Franka performing various manipulation tasks
""",
homepage='https://github.com/haosulab/ManiSkill2',
data_path='gs://gresearch/robotics/maniskill_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=151.05 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'episode_id': Text(shape=(), dtype=string),
'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({
'base_pose': Tensor(shape=(7,), dtype=float32),
'depth': Image(shape=(256, 256, 1), dtype=uint16),
'image': Image(shape=(256, 256, 3), dtype=uint8),
'main_camera_cam2world_gl': Tensor(shape=(4, 4), dtype=float32),
'main_camera_extrinsic_cv': Tensor(shape=(4, 4), dtype=float32),
'main_camera_intrinsic_cv': Tensor(shape=(3, 3), dtype=float32),
'state': Tensor(shape=(18,), dtype=float32),
'target_object_or_part_final_pose': Tensor(shape=(7,), dtype=float32),
'target_object_or_part_final_pose_valid': Tensor(shape=(7,), dtype=uint8),
'target_object_or_part_initial_pose': Tensor(shape=(7,), dtype=float32),
'target_object_or_part_initial_pose_valid': Tensor(shape=(7,), dtype=uint8),
'tcp_pose': Tensor(shape=(7,), dtype=float32),
'wrist_camera_cam2world_gl': Tensor(shape=(4, 4), dtype=float32),
'wrist_camera_extrinsic_cv': Tensor(shape=(4, 4), dtype=float32),
'wrist_camera_intrinsic_cv': Tensor(shape=(3, 3), dtype=float32),
'wrist_depth': Image(shape=(256, 256, 1), dtype=uint16),
'wrist_image': Image(shape=(256, 256, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{gu2023maniskill2,
title={ManiSkill2: A Unified Benchmark for Generalizable Manipulation Skills},
author={Gu, Jiayuan and Xiang, Fanbo and Li, Xuanlin and Ling, Zhan and Liu, Xiqiang and Mu, Tongzhou and Tang, Yihe and Tao, Stone and Wei, Xinyue and Yao, Yunchao and Yuan, Xiaodi and Xie, Pengwei and Huang, Zhiao and Chen, Rui and Su, Hao},
booktitle={International Conference on Learning Representations},
year={2023}
}""",
)
| highres_image (480, 640, 3) | image (64, 64, 3) |
|---|---|
tfds.core.DatasetInfo(
name='cmu_franka_exploration_dataset_converted_externally_to_rlds',
full_name='cmu_franka_exploration_dataset_converted_externally_to_rlds/0.1.0',
description="""
Franka exploring toy kitchens
""",
homepage='https://human-world-model.github.io/',
data_path='gs://gresearch/robotics/cmu_franka_exploration_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=602.24 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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({
'highres_image': Image(shape=(480, 640, 3), dtype=uint8),
'image': Image(shape=(64, 64, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
'structured_action': Tensor(shape=(8,), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{mendonca2023structured,
title={Structured World Models from Human Videos},
author={Mendonca, Russell and Bahl, Shikhar and Pathak, Deepak},
journal={RSS},
year={2023}
}""",
)
| image (480, 640, 3) |
|---|
tfds.core.DatasetInfo(
name='ucsd_kitchen_dataset_converted_externally_to_rlds',
full_name='ucsd_kitchen_dataset_converted_externally_to_rlds/0.1.0',
description="""
xArm interacting with different toy kitchens
""",
homepage=' ',
data_path='gs://gresearch/robotics/ucsd_kitchen_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=1.33 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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=(480, 640, 3), dtype=uint8),
'state': Tensor(shape=(21,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@ARTICLE{ucsd_kitchens,
author = {Ge Yan, Kris Wu, and Xiaolong Wang},
title = {{ucsd kitchens Dataset}},
year = {2023},
month = {August}
}""",
)
| image (224, 224, 3) |
|---|
tfds.core.DatasetInfo(
name='ucsd_pick_and_place_dataset_converted_externally_to_rlds',
full_name='ucsd_pick_and_place_dataset_converted_externally_to_rlds/0.1.0',
description="""
xArm picking and placing objects with distractors
""",
homepage='https://owmcorl.github.io',
data_path='gs://gresearch/robotics/ucsd_pick_and_place_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=3.53 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'disclaimer': Text(shape=(), dtype=string),
'file_path': Text(shape=(), dtype=string),
'n_transitions': Scalar(shape=(), dtype=int32),
'success': Scalar(shape=(), dtype=bool),
'success_labeled_by': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(4,), 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=(224, 224, 3), dtype=uint8),
'state': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@preprint{Feng2023Finetuning,
title={Finetuning Offline World Models in the Real World},
author={Yunhai Feng, Nicklas Hansen, Ziyan Xiong, Chandramouli Rajagopalan, Xiaolong Wang},
year={2023}
}""",
)
| wrist_image (128, 128, 3) | image (128, 128, 3) |
|---|---|
tfds.core.DatasetInfo(
name='austin_sailor_dataset_converted_externally_to_rlds',
full_name='austin_sailor_dataset_converted_externally_to_rlds/0.1.0',
description="""
Franka tablesetting tasks
""",
homepage='https://ut-austin-rpl.github.io/sailor/',
data_path='gs://gresearch/robotics/austin_sailor_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=18.85 GiB,
features=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=(128, 128, 3), dtype=uint8),
'state': Tensor(shape=(8,), dtype=float32),
'state_ee': Tensor(shape=(16,), dtype=float32),
'state_gripper': Tensor(shape=(1,), dtype=float32),
'state_joint': Tensor(shape=(7,), dtype=float32),
'wrist_image': Image(shape=(128, 128, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{nasiriany2022sailor,
title={Learning and Retrieval from Prior Data for Skill-based Imitation Learning},
author={Soroush Nasiriany and Tian Gao and Ajay Mandlekar and Yuke Zhu},
booktitle={Conference on Robot Learning (CoRL)},
year={2022}
}""",
)
| wrist_image (84, 84, 3) | image (84, 84, 3) |
|---|---|
tfds.core.DatasetInfo(
name='austin_sirius_dataset_converted_externally_to_rlds',
full_name='austin_sirius_dataset_converted_externally_to_rlds/0.1.0',
description="""
Franka tabletop manipulation tasks
""",
homepage='https://ut-austin-rpl.github.io/sirius/',
data_path='gs://gresearch/robotics/austin_sirius_dataset_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=6.55 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(7,), dtype=float32),
'action_mode': Tensor(shape=(1,), dtype=float32),
'discount': Scalar(shape=(), dtype=float32),
'intv_label': Tensor(shape=(1,), 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=(84, 84, 3), dtype=uint8),
'state': Tensor(shape=(8,), dtype=float32),
'state_ee': Tensor(shape=(16,), dtype=float32),
'state_gripper': Tensor(shape=(1,), dtype=float32),
'state_joint': Tensor(shape=(7,), dtype=float32),
'wrist_image': Image(shape=(84, 84, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{liu2022robot,
title = {Robot Learning on the Job: Human-in-the-Loop Autonomy and Learning During Deployment},
author = {Huihan Liu and Soroush Nasiriany and Lance Zhang and Zhiyao Bao and Yuke Zhu},
booktitle = {Robotics: Science and Systems (RSS)},
year = {2023}
}""",
)
| natural_language_instruction | image (171, 213, 3) |
|---|---|
| place the white sponge in the ceramic bowl | |
| place the eraser on the white sponge | |
| move the arm in a circular motion |
tfds.core.DatasetInfo(
name='bc_z',
full_name='bc_z/0.1.0',
description="""
Teleoped Google robot doing mostly pick-place from a table
""",
homepage='https://www.kaggle.com/datasets/google/bc-z-robot/discussion/309201',
data_path='gs://gresearch/robotics/bc_z/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=80.54 GiB,
features=FeaturesDict({
'steps': Dataset({
'action': FeaturesDict({
'future/axis_angle_residual': Tensor(shape=(30,), dtype=float32),
'future/target_close': Tensor(shape=(10,), dtype=int64),
'future/xyz_residual': Tensor(shape=(30,), dtype=float32),
}),
'is_first': bool,
'is_last': bool,
'is_terminal': bool,
'observation': FeaturesDict({
'episode_success': float32,
'image': Image(shape=(171, 213, 3), dtype=uint8),
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
'natural_language_instruction': string,
'present/autonomous': int64,
'present/axis_angle': Tensor(shape=(3,), dtype=float32),
'present/intervention': int64,
'present/sensed_close': Tensor(shape=(1,), dtype=float32),
'present/xyz': Tensor(shape=(3,), dtype=float32),
'sequence_length': int64,
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""@inproceedings{jang2021bc,
title={{BC}-Z: Zero-Shot Task Generalization with Robotic Imitation Learning},
author={Eric Jang and Alex Irpan and Mohi Khansari and Daniel Kappler and Frederik Ebert and Corey Lynch and Sergey Levine and Chelsea Finn},
booktitle={5th Annual Conference on Robot Learning},
year={2021},
url={https://openreview.net/forum?id=8kbp23tSGYv}}""",
)
| image (32, 32, 3) |
|---|
tfds.core.DatasetInfo(
name='usc_cloth_sim_converted_externally_to_rlds',
full_name='usc_cloth_sim_converted_externally_to_rlds/0.1.0',
description="""
Franka cloth interaction tasks
""",
homepage='https://uscresl.github.io/dmfd/',
data_path='gs://gresearch/robotics/usc_cloth_sim_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=254.52 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(4,), 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=(32, 32, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""@article{salhotra2022dmfd,
author={Salhotra, Gautam and Liu, I-Chun Arthur and Dominguez-Kuhne, Marcus and Sukhatme, Gaurav S.},
journal={IEEE Robotics and Automation Letters},
title={Learning Deformable Object Manipulation From Expert Demonstrations},
year={2022},
volume={7},
number={4},
pages={8775-8782},
doi={10.1109/LRA.2022.3187843}
}""",
)
| image (128, 128, 3) |
|---|
tfds.core.DatasetInfo(
name='utokyo_pr2_opening_fridge_converted_externally_to_rlds',
full_name='utokyo_pr2_opening_fridge_converted_externally_to_rlds/0.1.0',
description="""
PR2 opening fridge doors
""",
homepage='--',
data_path='gs://gresearch/robotics/utokyo_pr2_opening_fridge_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=360.57 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""@misc{oh2023pr2utokyodatasets,
author={Jihoon Oh and Naoaki Kanazawa and Kento Kawaharazuka},
title={X-Embodiment U-Tokyo PR2 Datasets},
year={2023},
url={https://github.com/ojh6404/rlds_dataset_builder},
}""",
)
| image (128, 128, 3) |
|---|
tfds.core.DatasetInfo(
name='utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds',
full_name='utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds/0.1.0',
description="""
tbd
""",
homepage='tbd',
data_path='gs://gresearch/robotics/utokyo_pr2_tabletop_manipulation_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=829.37 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""tbd""",
)
| wrist_image (64, 64, 3) | image (64, 64, 3) |
|---|---|
tfds.core.DatasetInfo(
name='utokyo_saytap_converted_externally_to_rlds',
full_name='utokyo_saytap_converted_externally_to_rlds/0.1.0',
description="""
A1 walking, no RGB
""",
homepage='https://saytap.github.io/',
data_path='gs://gresearch/robotics/utokyo_saytap_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=55.34 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(12,), 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({
'desired_pattern': Tensor(shape=(4, 5), dtype=bool),
'desired_vel': Tensor(shape=(3,), dtype=float32),
'image': Image(shape=(64, 64, 3), dtype=uint8),
'prev_act': Tensor(shape=(12,), dtype=float32),
'proj_grav_vec': Tensor(shape=(3,), dtype=float32),
'state': Tensor(shape=(30,), dtype=float32),
'wrist_image': Image(shape=(64, 64, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{saytap2023,
author = {Yujin Tang and Wenhao Yu and Jie Tan and Heiga Zen and Aleksandra Faust and
Tatsuya Harada},
title = {SayTap: Language to Quadrupedal Locomotion},
eprint = {arXiv:2306.07580},
url = {https://saytap.github.io},
note = "{https://saytap.github.io}",
year = {2023}
}""",
)
| image (224, 224, 3) | hand_image (224, 224, 3) | image2 (224, 224, 3) |
|---|---|---|
tfds.core.DatasetInfo(
name='utokyo_xarm_pick_and_place_converted_externally_to_rlds',
full_name='utokyo_xarm_pick_and_place_converted_externally_to_rlds/0.1.0',
description="""
xArm picking and placing objects
""",
homepage='--',
data_path='gs://gresearch/robotics/utokyo_xarm_pick_and_place_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=1.29 GiB,
features=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({
'end_effector_pose': Tensor(shape=(6,), dtype=float32),
'hand_image': Image(shape=(224, 224, 3), dtype=uint8),
'image': Image(shape=(224, 224, 3), dtype=uint8),
'image2': Image(shape=(224, 224, 3), dtype=uint8),
'joint_state': Tensor(shape=(14,), dtype=float32),
'joint_trajectory': Tensor(shape=(21,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""@misc{matsushima2023weblab,
title={Weblab xArm Dataset},
author={Tatsuya Matsushima and Hiroki Furuta and Yusuke Iwasawa and Yutaka Matsuo},
year={2023},
}""",
)
| image (256, 256, 3) |
|---|
tfds.core.DatasetInfo(
name='utokyo_xarm_bimanual_converted_externally_to_rlds',
full_name='utokyo_xarm_bimanual_converted_externally_to_rlds/0.1.0',
description="""
xArm bimanual setup folding towel
""",
homepage='--',
data_path='gs://gresearch/robotics/utokyo_xarm_bimanual_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=138.44 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(14,), 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({
'action_l': Tensor(shape=(7,), dtype=float32),
'action_r': Tensor(shape=(7,), dtype=float32),
'image': Image(shape=(256, 256, 3), dtype=uint8),
'pose_l': Tensor(shape=(6,), dtype=float32),
'pose_r': Tensor(shape=(6,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""@misc{matsushima2023weblab,
title={Weblab xArm Dataset},
author={Tatsuya Matsushima and Hiroki Furuta and Yusuke Iwasawa and Yutaka Matsuo},
year={2023},
}""",
)
| image2 (240, 320, 3) | image (240, 320, 3) | image1 (240, 320, 3) |
|---|---|---|
tfds.core.DatasetInfo(
name='robo_net',
full_name='robo_net/1.0.0',
description="""
This is an excerpt of the [RoboNet](https://github.com/SudeepDasari/RoboNet) dataset.
Data from 5 robots randomly interacting with a bin using the AutoGrasp primitive. The action/state space is shared across all robots, and camera observations were taken using 3 synced camera images.
""",
homepage='https://www.tensorflow.org/datasets/catalog/robo_net',
data_path='gs://gresearch/robotics/robo_net/1.0.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=799.91 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
'robot': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(5,), dtype=float32),
'discount': Scalar(shape=(), dtype=float32),
'is_first': Scalar(shape=(), dtype=bool),
'is_last': Scalar(shape=(), dtype=bool),
'is_terminal': Scalar(shape=(), dtype=bool),
'language_embedding': Tensor(shape=(512,), dtype=float32),
'language_instruction': Text(shape=(), dtype=string),
'observation': FeaturesDict({
'image': Image(shape=(240, 320, 3), dtype=uint8),
'image1': Image(shape=(240, 320, 3), dtype=uint8),
'image2': Image(shape=(240, 320, 3), dtype=uint8),
'state': Tensor(shape=(5,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{dasari2019robonet,
title={RoboNet: Large-Scale Multi-Robot Learning},
author={Sudeep Dasari and Frederik Ebert and Stephen Tian and Suraj Nair and Bernadette Bucher and Karl Schmeckpeper and Siddharth Singh and Sergey Levine and Chelsea Finn},
year={2019},
eprint={1910.11215},
archivePrefix={arXiv},
primaryClass={cs.RO},
booktitle={CoRL 2019: Volume 100 Proceedings of Machine Learning Research}
}""",
)
| hand_image (480, 640, 3) |
|---|
tfds.core.DatasetInfo(
name='berkeley_mvp_converted_externally_to_rlds',
full_name='berkeley_mvp_converted_externally_to_rlds/0.1.0',
description="""
xArm performing 6 manipulation tasks
""",
homepage='https://arxiv.org/abs/2203.06173',
data_path='gs://gresearch/robotics/berkeley_mvp_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=12.34 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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({
'gripper': Scalar(shape=(), dtype=bool),
'hand_image': Image(shape=(480, 640, 3), dtype=uint8),
'joint_pos': Tensor(shape=(7,), dtype=float32),
'pose': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@InProceedings{Radosavovic2022,
title = {Real-World Robot Learning with Masked Visual Pre-training},
author = {Ilija Radosavovic and Tete Xiao and Stephen James and Pieter Abbeel and Jitendra Malik and Trevor Darrell},
booktitle = {CoRL},
year = {2022}
}""",
)
| hand_image (480, 640, 3) |
|---|
tfds.core.DatasetInfo(
name='berkeley_rpt_converted_externally_to_rlds',
full_name='berkeley_rpt_converted_externally_to_rlds/0.1.0',
description="""
Franka performing tabletop pick place tasks
""",
homepage='https://arxiv.org/abs/2306.10007',
data_path='gs://gresearch/robotics/berkeley_rpt_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=40.64 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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({
'gripper': Scalar(shape=(), dtype=bool),
'hand_image': Image(shape=(480, 640, 3), dtype=uint8),
'joint_pos': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{Radosavovic2023,
title={Robot Learning with Sensorimotor Pre-training},
author={Ilija Radosavovic and Baifeng Shi and Letian Fu and Ken Goldberg and Trevor Darrell and Jitendra Malik},
year={2023},
journal={arXiv:2306.10007}
}""",
)
| image (480, 640, 3) |
|---|
tfds.core.DatasetInfo(
name='kaist_nonprehensile_converted_externally_to_rlds',
full_name='kaist_nonprehensile_converted_externally_to_rlds/0.1.0',
description="""
Franka manipulating ungraspable objects
""",
homepage='--',
data_path='gs://gresearch/robotics/kaist_nonprehensile_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=11.71 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(20,), 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=(480, 640, 3), dtype=uint8),
'partial_pointcloud': Tensor(shape=(512, 3), dtype=float32),
'state': Tensor(shape=(21,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{kimpre,
title={Pre-and post-contact policy decomposition for non-prehensile manipulation with zero-shot sim-to-real transfer},
author={Kim, Minchan and Han, Junhyek and Kim, Jaehyung and Kim, Beomjoon},
booktitle={2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
year={2023},
organization={IEEE}
}""",
)
| image (480, 480, 3) |
|---|
tfds.core.DatasetInfo(
name='stanford_mask_vit_converted_externally_to_rlds',
full_name='stanford_mask_vit_converted_externally_to_rlds/0.1.0',
description="""
Sawyer pushing and picking objects in a bin
""",
homepage='https://arxiv.org/abs/2206.11894',
data_path='gs://gresearch/robotics/stanford_mask_vit_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=76.17 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(5,), 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({
'end_effector_pose': Tensor(shape=(5,), dtype=float32),
'finger_sensors': Tensor(shape=(1,), dtype=float32),
'high_bound': Tensor(shape=(5,), dtype=float32),
'image': Image(shape=(480, 480, 3), dtype=uint8),
'low_bound': Tensor(shape=(5,), dtype=float32),
'state': Tensor(shape=(15,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
'val': ,
},
citation="""@inproceedings{gupta2022maskvit,
title={MaskViT: Masked Visual Pre-Training for Video Prediction},
author={Agrim Gupta and Stephen Tian and Yunzhi Zhang and Jiajun Wu and Roberto Martín-Martín and Li Fei-Fei},
booktitle={International Conference on Learning Representations},
year={2022}
}""",
)
| image (120, 120, 3) |
|---|
tfds.core.DatasetInfo(
name='tokyo_u_lsmo_converted_externally_to_rlds',
full_name='tokyo_u_lsmo_converted_externally_to_rlds/0.1.0',
description="""
motion planning trajectory of pick place tasks
""",
homepage='https://journals.sagepub.com/doi/full/10.1177/02783649211044405',
data_path='gs://gresearch/robotics/tokyo_u_lsmo_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=335.71 MiB,
features=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),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@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},
}""",
)
| image (480, 640, 3) |
|---|
tfds.core.DatasetInfo(
name='dlr_sara_pour_converted_externally_to_rlds',
full_name='dlr_sara_pour_converted_externally_to_rlds/0.1.0',
description="""
pouring liquid from a bottle into a mug
""",
homepage='https://elib.dlr.de/193739/1/padalkar2023rlsct.pdf',
data_path='gs://gresearch/robotics/dlr_sara_pour_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=2.92 GiB,
features=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=(480, 640, 3), dtype=uint8),
'state': Tensor(shape=(6,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{padalkar2023guiding,
title={Guiding Reinforcement Learning with Shared Control Templates},
author={Padalkar, Abhishek and Quere, Gabriel and Steinmetz, Franz and Raffin, Antonin and Nieuwenhuisen, Matthias and Silv{'e}rio, Jo{\~a}o and Stulp, Freek},
booktitle={40th IEEE International Conference on Robotics and Automation, ICRA 2023},
year={2023},
organization={IEEE}
}""",
)
| image (480, 640, 3) |
|---|
tfds.core.DatasetInfo(
name='dlr_sara_grid_clamp_converted_externally_to_rlds',
full_name='dlr_sara_grid_clamp_converted_externally_to_rlds/0.1.0',
description="""
place grid clamp onto grids on table
""",
homepage='https://www.researchsquare.com/article/rs-3289569/v1',
data_path='gs://gresearch/robotics/dlr_sara_grid_clamp_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=1.65 GiB,
features=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=(480, 640, 3), dtype=uint8),
'state': Tensor(shape=(12,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{padalkar2023guided,
title={A guided reinforcement learning approach using shared control templates for learning manipulation skills in the real world},
author={Padalkar, Abhishek and Quere, Gabriel and Raffin, Antonin and Silv{'e}rio, Jo{\~a}o and Stulp, Freek},
journal={Research square preprint rs-3289569/v1},
year={2023}
}""",
)
| image (360, 640, 3) |
|---|
| image (224, 224, 3) |
|---|
tfds.core.DatasetInfo(
name='asu_table_top_converted_externally_to_rlds',
full_name='asu_table_top_converted_externally_to_rlds/0.1.0',
description="""
UR5 performing table-top pick/place/rotate tasks
""",
homepage='https://link.springer.com/article/10.1007/s10514-023-10129-1',
data_path='gs://gresearch/robotics/asu_table_top_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=737.60 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(7,), dtype=float32),
'action_delta': Tensor(shape=(7,), dtype=float32),
'action_inst': Text(shape=(), dtype=string),
'discount': Scalar(shape=(), dtype=float32),
'goal_object': Text(shape=(), dtype=string),
'ground_truth_states': FeaturesDict({
'EE': Tensor(shape=(6,), dtype=float32),
'bottle': Tensor(shape=(6,), dtype=float32),
'bread': Tensor(shape=(6,), dtype=float32),
'coke': Tensor(shape=(6,), dtype=float32),
'cube': Tensor(shape=(6,), dtype=float32),
'milk': Tensor(shape=(6,), dtype=float32),
'pepsi': Tensor(shape=(6,), 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=(224, 224, 3), dtype=uint8),
'state': Tensor(shape=(7,), dtype=float32),
'state_vel': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
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}
}""",
)
| image_3 (256, 256, 3) | image_2 (256, 256, 3) | image_4 (256, 256, 3) | image_1 (256, 256, 3) |
|---|---|---|---|
tfds.core.DatasetInfo(
name='stanford_robocook_converted_externally_to_rlds',
full_name='stanford_robocook_converted_externally_to_rlds/0.1.0',
description="""
Franka preparing dumplings with various tools
""",
homepage='https://hshi74.github.io/robocook/',
data_path='gs://gresearch/robotics/stanford_robocook_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=124.62 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'extrinsics_1': Tensor(shape=(4, 4), dtype=float32),
'extrinsics_2': Tensor(shape=(4, 4), dtype=float32),
'extrinsics_3': Tensor(shape=(4, 4), dtype=float32),
'extrinsics_4': Tensor(shape=(4, 4), dtype=float32),
'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({
'depth_1': Tensor(shape=(256, 256), dtype=float32),
'depth_2': Tensor(shape=(256, 256), dtype=float32),
'depth_3': Tensor(shape=(256, 256), dtype=float32),
'depth_4': Tensor(shape=(256, 256), dtype=float32),
'image_1': Image(shape=(256, 256, 3), dtype=uint8),
'image_2': Image(shape=(256, 256, 3), dtype=uint8),
'image_3': Image(shape=(256, 256, 3), dtype=uint8),
'image_4': Image(shape=(256, 256, 3), dtype=uint8),
'state': Tensor(shape=(7,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{shi2023robocook,
title={RoboCook: Long-Horizon Elasto-Plastic Object Manipulation with Diverse Tools},
author={Shi, Haochen and Xu, Huazhe and Clarke, Samuel and Li, Yunzhu and Wu, Jiajun},
journal={arXiv preprint arXiv:2306.14447},
year={2023}
}""",
)
| image (64, 64, 3) |
|---|
tfds.core.DatasetInfo(
name='eth_agent_affordances',
full_name='eth_agent_affordances/0.1.0',
description="""
Franka opening ovens -- point cloud + proprio only
""",
homepage='https://ieeexplore.ieee.org/iel7/10160211/10160212/10160747.pdf',
data_path='gs://gresearch/robotics/eth_agent_affordances/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=17.27 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
'input_point_cloud': Tensor(shape=(10000, 3), dtype=float16),
}),
'steps': Dataset({
'action': Tensor(shape=(6,), 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=(64, 64, 3), dtype=uint8),
'input_point_cloud': Tensor(shape=(10000, 3), dtype=float16),
'state': Tensor(shape=(8,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{schiavi2023learning,
title={Learning agent-aware affordances for closed-loop interaction with articulated objects},
author={Schiavi, Giulio and Wulkop, Paula and Rizzi, Giuseppe and Ott, Lionel and Siegwart, Roland and Chung, Jen Jen},
booktitle={2023 IEEE International Conference on Robotics and Automation (ICRA)},
pages={5916--5922},
year={2023},
organization={IEEE}
}""",
)
| wrist_image (64, 64, 3) | image (64, 64, 3) |
|---|---|
tfds.core.DatasetInfo(
name='imperialcollege_sawyer_wrist_cam',
full_name='imperialcollege_sawyer_wrist_cam/0.1.0',
description="""
Sawyer performing table top manipulation
""",
homepage='--',
data_path='gs://gresearch/robotics/imperialcollege_sawyer_wrist_cam/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=81.87 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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=(64, 64, 3), dtype=uint8),
'state': Tensor(shape=(1,), dtype=float32),
'wrist_image': Image(shape=(64, 64, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""--""",
)
| image (360, 640, 3) | wrist_image (240, 320, 3) |
|---|---|
tfds.core.DatasetInfo(
name='iamlab_cmu_pickup_insert_converted_externally_to_rlds',
full_name='iamlab_cmu_pickup_insert_converted_externally_to_rlds/0.1.0',
description="""
Franka picking objects and insertion tasks
""",
homepage='https://openreview.net/forum?id=WuBv9-IGDUA',
data_path='gs://gresearch/robotics/iamlab_cmu_pickup_insert_converted_externally_to_rlds/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=50.29 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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=(360, 640, 3), dtype=uint8),
'state': Tensor(shape=(20,), dtype=float32),
'wrist_image': Image(shape=(240, 320, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{
saxena2023multiresolution,
title={Multi-Resolution Sensing for Real-Time Control with Vision-Language Models},
author={Saumya Saxena and Mohit Sharma and Oliver Kroemer},
booktitle={7th Annual Conference on Robot Learning},
year={2023},
url={https://openreview.net/forum?id=WuBv9-IGDUA}
}""",
)
| image_1 (360, 640, 3) | image_3 (360, 640, 3) | image_2 (360, 640, 3) | image_4 (360, 640, 3) |
|---|---|---|---|
tfds.core.DatasetInfo(
name='uiuc_d3field',
full_name='uiuc_d3field/0.1.0',
description="""
Organizing office desk, utensils etc
""",
homepage='https://robopil.github.io/d3fields/',
data_path='gs://gresearch/robotics/uiuc_d3field/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=15.82 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(3,), 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_1': Image(shape=(360, 640, 1), dtype=uint16),
'depth_2': Image(shape=(360, 640, 1), dtype=uint16),
'depth_3': Image(shape=(360, 640, 1), dtype=uint16),
'depth_4': Image(shape=(360, 640, 1), dtype=uint16),
'image_1': Image(shape=(360, 640, 3), dtype=uint8),
'image_2': Image(shape=(360, 640, 3), dtype=uint8),
'image_3': Image(shape=(360, 640, 3), dtype=uint8),
'image_4': Image(shape=(360, 640, 3), dtype=uint8),
'state': Tensor(shape=(4, 4), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{wang2023d3field,
title={D^3Field: Dynamic 3D Descriptor Fields for Generalizable Robotic Manipulation},
author={Wang, Yixuan and Li, Zhuoran and Zhang, Mingtong and Driggs-Campbell, Katherine and Wu, Jiajun and Fei-Fei, Li and Li, Yunzhu},
journal={arXiv preprint arXiv:},
year={2023},
}""",
)
| image (128, 128, 3) | wrist_image (128, 128, 3) |
|---|---|
tfds.core.DatasetInfo(
name='utaustin_mutex',
full_name='utaustin_mutex/0.1.0',
description="""
Diverse household manipulation tasks
""",
homepage='https://ut-austin-rpl.github.io/MUTEX/',
data_path='gs://gresearch/robotics/utaustin_mutex/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=20.79 GiB,
features=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=(128, 128, 3), dtype=uint8),
'state': Tensor(shape=(24,), dtype=float32),
'wrist_image': Image(shape=(128, 128, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{
shah2023mutex,
title={{MUTEX}: Learning Unified Policies from Multimodal Task Specifications},
author={Rutav Shah and Roberto Mart{'\i}n-Mart{'\i}n and Yuke Zhu},
booktitle={7th Annual Conference on Robot Learning},
year={2023},
url={https://openreview.net/forum?id=PwqiqaaEzJ}
}""",
)
| wrist_image (224, 224, 3) | image (224, 224, 3) |
|---|---|
tfds.core.DatasetInfo(
name='berkeley_fanuc_manipulation',
full_name='berkeley_fanuc_manipulation/0.1.0',
description="""
Fanuc robot performing various manipulation tasks
""",
homepage='https://sites.google.com/berkeley.edu/fanuc-manipulation',
data_path='gs://gresearch/robotics/berkeley_fanuc_manipulation/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=8.85 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(6,), 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({
'end_effector_state': Tensor(shape=(7,), dtype=float32),
'image': Image(shape=(224, 224, 3), dtype=uint8),
'state': Tensor(shape=(13,), dtype=float32),
'wrist_image': Image(shape=(224, 224, 3), dtype=uint8),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{fanuc_manipulation2023,
title={Fanuc Manipulation: A Dataset for Learning-based Manipulation with FANUC Mate 200iD Robot},
author={Zhu, Xinghao and Tian, Ran and Xu, Chenfeng and Ding, Mingyu and Zhan, Wei and Tomizuka, Masayoshi},
year={2023},
}""",
)
| image (128, 128, 3) |
|---|
tfds.core.DatasetInfo(
name='cmu_play_fusion',
full_name='cmu_play_fusion/0.1.0',
description="""
The robot plays with 3 complex scenes: a grill with many cooking objects like toaster, pan, etc. It has to pick, open, place, close. It has to set a table, move plates, cups, utensils. And it has to place dishes in the sink, dishwasher, hand cups etc.
""",
homepage='https://play-fusion.github.io/',
data_path='gs://gresearch/robotics/cmu_play_fusion/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=6.68 GiB,
features=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),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@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}
}""",
)
| image (128, 128, 3) |
|---|
tfds.core.DatasetInfo(
name='cmu_stretch',
full_name='cmu_stretch/0.1.0',
description="""
Hello stretch robot kitchen interactions
""",
homepage='https://robo-affordances.github.io/',
data_path='gs://gresearch/robotics/cmu_stretch/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=728.06 MiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(8,), 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=(4,), dtype=float32),
}),
'reward': Scalar(shape=(), dtype=float32),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{bahl2023affordances,
title={Affordances from Human Videos as a Versatile Representation for Robotics},
author={Bahl, Shikhar and Mendonca, Russell and Chen, Lili and Jain, Unnat and Pathak, Deepak},
booktitle={CVPR},
year={2023}
}
@article{mendonca2023structured,
title={Structured World Models from Human Videos},
author={Mendonca, Russell and Bahl, Shikhar and Pathak, Deepak},
journal={CoRL},
year={2023}
}""",
)
| image (120, 160, 3) |
|---|
tfds.core.DatasetInfo(
name='berkeley_gnm_recon',
full_name='berkeley_gnm_recon/0.1.0',
description="""
off-road navigation
""",
homepage='https://sites.google.com/view/recon-robot',
data_path='gs://gresearch/robotics/berkeley_gnm_recon/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=18.73 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(2,), dtype=float64),
'action_angle': Tensor(shape=(3,), dtype=float64),
'discount': Scalar(shape=(), dtype=float64),
'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, 160, 3), dtype=uint8),
'position': Tensor(shape=(2,), dtype=float64),
'state': Tensor(shape=(3,), dtype=float64),
'yaw': Tensor(shape=(1,), dtype=float64),
}),
'reward': Scalar(shape=(), dtype=float64),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{
shah2021rapid,
title={{Rapid Exploration for Open-World Navigation with Latent Goal Models}},
author={Dhruv Shah and Benjamin Eysenbach and Nicholas Rhinehart and Sergey Levine},
booktitle={5th Annual Conference on Robot Learning },
year={2021},
url={https://openreview.net/forum?id=d_SWJhyKfVw}
}""",
)
| image (64, 85, 3) |
|---|
tfds.core.DatasetInfo(
name='berkeley_gnm_cory_hall',
full_name='berkeley_gnm_cory_hall/0.1.0',
description="""
hallway navigation
""",
homepage='https://arxiv.org/abs/1709.10489',
data_path='gs://gresearch/robotics/berkeley_gnm_cory_hall/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=1.39 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(2,), dtype=float64),
'action_angle': Tensor(shape=(3,), dtype=float64),
'discount': Scalar(shape=(), dtype=float64),
'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=(64, 85, 3), dtype=uint8),
'position': Tensor(shape=(2,), dtype=float64),
'state': Tensor(shape=(3,), dtype=float64),
'yaw': Tensor(shape=(1,), dtype=float64),
}),
'reward': Scalar(shape=(), dtype=float64),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@inproceedings{kahn2018self,
title={Self-supervised deep reinforcement learning with generalized computation graphs for robot navigation},
author={Kahn, Gregory and Villaflor, Adam and Ding, Bosen and Abbeel, Pieter and Levine, Sergey},
booktitle={2018 IEEE international conference on robotics and automation (ICRA)},
pages={5129--5136},
year={2018},
organization={IEEE}
}""",
)
| image (120, 160, 3) |
|---|
tfds.core.DatasetInfo(
name='berkeley_gnm_sac_son',
full_name='berkeley_gnm_sac_son/0.1.0',
description="""
office navigation
""",
homepage='https://sites.google.com/view/SACSoN-review',
data_path='gs://gresearch/robotics/berkeley_gnm_sac_son/0.1.0',
file_format=tfrecord,
download_size=Unknown size,
dataset_size=7.00 GiB,
features=FeaturesDict({
'episode_metadata': FeaturesDict({
'file_path': Text(shape=(), dtype=string),
}),
'steps': Dataset({
'action': Tensor(shape=(2,), dtype=float64),
'action_angle': Tensor(shape=(3,), dtype=float64),
'discount': Scalar(shape=(), dtype=float64),
'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, 160, 3), dtype=uint8),
'position': Tensor(shape=(2,), dtype=float64),
'state': Tensor(shape=(3,), dtype=float64),
'yaw': Tensor(shape=(1,), dtype=float64),
}),
'reward': Scalar(shape=(), dtype=float64),
}),
}),
supervised_keys=None,
disable_shuffling=False,
splits={
'train': ,
},
citation="""@article{hirose2023sacson,
title={SACSoN: Scalable Autonomous Data Collection for Social Navigation},
author={Hirose, Noriaki and Shah, Dhruv and Sridhar, Ajay and Levine, Sergey},
journal={arXiv preprint arXiv:2306.01874},
year={2023}
}""",
)