버클리_gnm_sac_son

  • 설명 :

사무실 내비게이션

나뉘다
'train' 2,955
  • 기능 구조 :
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),
    }),
})
  • 기능 문서 :
특징 수업 모양 Dtype 설명
특징Dict
에피소드_메타데이터 특징Dict
에피소드_메타데이터/파일_경로 텍스트 원본 데이터 파일의 경로입니다.
단계 데이터세트
단계/작업 텐서 (2,) float64 로봇 액션, 2x 위치로 구성
단계/action_angle 텐서 (삼,) float64 로봇 동작은 2x 위치, 1x 요로 구성됩니다.
걸음수/할인 스칼라 float64 할인이 제공되면 기본값은 1입니다.
걸음수/is_first 텐서 부울
걸음수/is_last 텐서 부울
단계/is_terminal 텐서 부울
단계/언어_임베딩 텐서 (512,) float32 코나 언어 임베딩. https://tfhub.dev/google/universal-sentence-encoder-large/5를 참조하세요.
단계/언어_지시 텍스트 언어 교육.
단계/관찰 특징Dict
단계/관찰/이미지 영상 (120, 160, 3) uint8 메인 카메라 RGB 관찰.
걸음 수/관찰/위치 텐서 (2,) float64 로봇 위치
단계/관찰/상태 텐서 (삼,) float64 로봇 상태는 [2x 위치, 1x 요]로 구성됩니다.
단계/관찰/요 텐서 (1,) float64 로봇 요
걸음수/보상 스칼라 float64 제공되는 경우 보상, 데모의 마지막 단계에서 1개.
  • 인용 :
@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}
}