robo_set

  • Keterangan :

Kumpulan data nyata dari satu lengan robot yang menunjukkan 12 keterampilan manipulasi non-sepele dalam 38 tugas, 7500 lintasan.

Membelah Contoh
'train' 18.250
  • Struktur fitur :
FeaturesDict({
    'episode_metadata': FeaturesDict({
        'file_path': string,
        'trial_id': string,
    }),
    'steps': Dataset({
        'action': Tensor(shape=(8,), dtype=float32),
        'discount': Scalar(shape=(), dtype=float32),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'language_instruction': string,
        'observation': FeaturesDict({
            'image_left': Image(shape=(240, 424, 3), dtype=uint8),
            'image_right': Image(shape=(240, 424, 3), dtype=uint8),
            'image_top': Image(shape=(240, 424, 3), dtype=uint8),
            'image_wrist': Image(shape=(240, 424, 3), dtype=uint8),
            'state': Tensor(shape=(8,), dtype=float32),
            'state_velocity': Tensor(shape=(8,), dtype=float32),
        }),
        'reward': Scalar(shape=(), dtype=float32),
    }),
})
  • Dokumentasi fitur :
Fitur Kelas Membentuk Tipe D Keterangan
FiturDict
episode_metadata FiturDict
episode_metadata/file_path Tensor rangkaian
episode_metadata/trial_id Tensor rangkaian
Langkah Himpunan data
langkah/tindakan Tensor (8,) float32
langkah/diskon Skalar float32
langkah/adalah_pertama Tensor bodoh
langkah/adalah_terakhir Tensor bodoh
langkah/is_terminal Tensor bodoh
langkah/bahasa_instruksi Tensor rangkaian
langkah/pengamatan FiturDict
langkah/pengamatan/gambar_kiri Gambar (240, 424, 3) uint8
langkah/pengamatan/image_right Gambar (240, 424, 3) uint8
langkah/pengamatan/image_top Gambar (240, 424, 3) uint8
langkah/pengamatan/image_wrist Gambar (240, 424, 3) uint8
langkah/pengamatan/keadaan Tensor (8,) float32
langkah/pengamatan/state_velocity Tensor (8,) float32
langkah/hadiah Skalar float32
@misc{bharadhwaj2023roboagent, title={RoboAgent: Generalization and Efficiency in Robot Manipulation via Semantic Augmentations and Action Chunking}, author={Homanga Bharadhwaj and Jay Vakil and Mohit Sharma and Abhinav Gupta and Shubham Tulsiani and Vikash Kumar},  year={2023}, eprint={2309.01918}, archivePrefix={arXiv}, primaryClass={cs.RO} }