mt_opt

  • 설명 :

MT-Opt 논문 용 데이터 세트 .

@misc{kalashnikov2021mtopt,
      title={MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale},
      author={Dmitry Kalashnikov and Jacob Varley and Yevgen Chebotar and Benjamin Swanson and Rico Jonschkowski and Chelsea Finn and Sergey Levine and Karol Hausman},
      year={2021},
      eprint={2104.08212},
      archivePrefix={arXiv},
      primaryClass={cs.RO}
}

mt_opt/rlds(기본 구성)

  • 구성 설명 : 이 데이터 세트에는 실제 로봇 전체에서 수집된 작업 에피소드가 포함되어 있습니다. RLDS 형식 을 따라 단계와 에피소드를 나타냅니다.

  • 데이터세트 크기 : 4.38 TiB

  • 분할 :

나뉘다
'train' 920,165
  • 기능 구조 :
FeaturesDict({
    'episode_id': string,
    'skill': uint8,
    'steps': Dataset({
        'action': FeaturesDict({
            'close_gripper': bool,
            'open_gripper': bool,
            'target_pose': Tensor(shape=(7,), dtype=float32),
            'terminate': bool,
        }),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': FeaturesDict({
            'gripper_closed': bool,
            'height_to_bottom': float32,
            'image': Image(shape=(512, 640, 3), dtype=uint8),
            'state_dense': Tensor(shape=(7,), dtype=float32),
        }),
    }),
    'task_code': string,
})
  • 기능 문서 :
특징 수업 모양 D타입 설명
풍모Dict
episode_id 텐서
기술 텐서 uint8
단계 데이터 세트
단계/액션 풍모Dict
단계/작업/close_gripper 텐서 부울
단계/작업/open_gripper 텐서 부울
단계/액션/target_pose 텐서 (7,) float32
단계/작업/종료 텐서 부울
단계/is_first 텐서 부울
단계/is_last 텐서 부울
단계/is_terminal 텐서 부울
단계/관찰 풍모Dict
단계/관찰/gripper_closed 텐서 부울
단계/관찰/height_to_bottom 텐서 float32
단계/관찰/이미지 영상 (512, 640, 3) uint8
단계/관측/state_dense 텐서 (7,) float32
task_code 텐서

mt_opt/sd

  • 구성 설명 : 작업 완료에 대해 사람이 선별한 정의를 포함하는 성공 감지기 데이터 세트입니다.

  • 데이터세트 크기 : 548.56 GiB

  • 분할 :

나뉘다
'test' 94,636
'train' 380,234
  • 기능 구조 :
FeaturesDict({
    'image_0': Image(shape=(512, 640, 3), dtype=uint8),
    'image_1': Image(shape=(480, 640, 3), dtype=uint8),
    'image_2': Image(shape=(480, 640, 3), dtype=uint8),
    'success': bool,
    'task_code': string,
})
  • 기능 문서 :
특징 수업 모양 D타입 설명
풍모Dict
이미지_0 영상 (512, 640, 3) uint8
이미지_1 영상 (480, 640, 3) uint8
이미지_2 영상 (480, 640, 3) uint8
성공 텐서 부울
task_code 텐서