robosuite_panda_pick_place_can

  • Descrizione :

Questi set di dati sono stati creati con l'ambiente PickPlaceCan del simulatore di braccio robotico robosuite . I set di dati umani sono stati registrati da un singolo operatore utilizzando RLDS Creator e un controller gamepad.

I set di dati sintetici sono stati registrati utilizzando la libreria EnvLogger .

I set di dati seguono il formato RLDS per rappresentare passaggi ed episodi.

Gli episodi sono composti da 400 passi. In ogni episodio, viene aggiunto un tag quando l'attività è completata, questo tag viene memorizzato come parte dei metadati del passaggio personalizzato.

Si noti che, a causa della dipendenza da EnvLogger, la generazione di questo set di dati è attualmente supportata solo in ambienti Linux.

@misc{ramos2021rlds,
      title={RLDS: an Ecosystem to Generate, Share and Use Datasets in Reinforcement Learning},
      author={Sabela Ramos and Sertan Girgin and Léonard Hussenot and Damien Vincent and Hanna Yakubovich and Daniel Toyama and Anita Gergely and Piotr Stanczyk and Raphael Marinier and Jeremiah Harmsen and Olivier Pietquin and Nikola Momchev},
      year={2021},
      eprint={2111.02767},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

robosuite_panda_pick_place_can/human_dc29b40a (configurazione predefinita)

  • Descrizione della configurazione : set di dati generato dall'uomo (50 episodi).

  • Pagina iniziale : https://github.com/google-research/rlds

  • Dimensione del download : 96.67 MiB

  • Dimensione del set di dati: 407.24 MiB

  • Divisioni :

Diviso Esempi
'train' 50
  • Struttura delle caratteristiche :
FeaturesDict({
    'agent_id': string,
    'episode_id': string,
    'episode_index': int32,
    'steps': Dataset({
        'action': Tensor(shape=(7,), dtype=float64),
        'discount': float64,
        'image': Image(shape=(None, None, 3), dtype=uint8),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': FeaturesDict({
            'Can_pos': Tensor(shape=(3,), dtype=float64),
            'Can_quat': Tensor(shape=(4,), dtype=float64),
            'Can_to_robot0_eef_pos': Tensor(shape=(3,), dtype=float64),
            'Can_to_robot0_eef_quat': Tensor(shape=(4,), dtype=float32),
            'object-state': Tensor(shape=(14,), dtype=float64),
            'robot0_eef_pos': Tensor(shape=(3,), dtype=float64),
            'robot0_eef_quat': Tensor(shape=(4,), dtype=float64),
            'robot0_gripper_qpos': Tensor(shape=(2,), dtype=float64),
            'robot0_gripper_qvel': Tensor(shape=(2,), dtype=float64),
            'robot0_joint_pos_cos': Tensor(shape=(7,), dtype=float64),
            'robot0_joint_pos_sin': Tensor(shape=(7,), dtype=float64),
            'robot0_joint_vel': Tensor(shape=(7,), dtype=float64),
            'robot0_proprio-state': Tensor(shape=(32,), dtype=float64),
        }),
        'reward': float64,
        'tag:placed': bool,
    }),
})
  • Documentazione delle funzionalità :
Caratteristica Classe Forma Tipo D Descrizione
CaratteristicheDict
ID_agente Tensore corda
episodio_id Tensore corda
indice_episodio Tensore int32
passi Set di dati
passi/azione Tensore (7,) galleggiante64
passi/sconto Tensore galleggiante64
passi/immagine Immagine (Nessuno, Nessuno, 3) uint8
passi/è_primo Tensore bool
passi/è_ultimo Tensore bool
passi/è_terminale Tensore bool
passi/osservazione CaratteristicheDict
passi/osservazione/Can_pos Tensore (3,) galleggiante64
passi/osservazione/Can_quat Tensore (4,) galleggiante64
passi/osservazione/Can_to_robot0_eef_pos Tensore (3,) galleggiante64
passi/osservazione/Can_to_robot0_eef_quat Tensore (4,) galleggiante32
passi/osservazione/stato-oggetto Tensore (14,) galleggiante64
passi/osservazione/robot0_eef_pos Tensore (3,) galleggiante64
passi/osservazione/robot0_eef_quat Tensore (4,) galleggiante64
passi/osservazione/robot0_gripper_qpos Tensore (2,) galleggiante64
passi/osservazione/robot0_gripper_qvel Tensore (2,) galleggiante64
passi/osservazione/robot0_joint_pos_cos Tensore (7,) galleggiante64
passi/osservazione/robot0_joint_pos_sin Tensore (7,) galleggiante64
passi/osservazione/robot0_joint_vel Tensore (7,) galleggiante64
steps/observation/robot0_proprio-state Tensore (32,) galleggiante64
passi/ricompensa Tensore galleggiante64
passi/tag:posto Tensore bool

robosuite_panda_pick_place_can/human_images_dc29b40a

  • Descrizione della configurazione : set di dati generato dall'uomo, comprese le immagini con diverse angolazioni della telecamera nell'osservazione. Tieni presente che la generazione potrebbe richiedere del tempo.

  • Pagina iniziale : https://www.tensorflow.org/datasets/catalog/robosuite_panda_pick_place_can

  • Dimensioni del download : 10.95 GiB

  • Dimensione del set di dati: 7.53 GiB

  • Divisioni :

Diviso Esempi
'train' 50
  • Struttura delle caratteristiche :
FeaturesDict({
    'agent_id': string,
    'episode_id': string,
    'episode_index': int32,
    'steps': Dataset({
        'action': Tensor(shape=(7,), dtype=float64),
        'discount': float64,
        'image': Image(shape=(None, None, 3), dtype=uint8),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': FeaturesDict({
            'Can_pos': Tensor(shape=(3,), dtype=float64),
            'Can_quat': Tensor(shape=(4,), dtype=float64),
            'Can_to_robot0_eef_pos': Tensor(shape=(3,), dtype=float64),
            'Can_to_robot0_eef_quat': Tensor(shape=(4,), dtype=float32),
            'agentview_image': Image(shape=(256, 256, 3), dtype=uint8),
            'birdview_image': Image(shape=(256, 256, 3), dtype=uint8),
            'object-state': Tensor(shape=(14,), dtype=float64),
            'robot0_eef_pos': Tensor(shape=(3,), dtype=float64),
            'robot0_eef_quat': Tensor(shape=(4,), dtype=float64),
            'robot0_eye_in_hand_image': Image(shape=(256, 256, 3), dtype=uint8),
            'robot0_gripper_qpos': Tensor(shape=(2,), dtype=float64),
            'robot0_gripper_qvel': Tensor(shape=(2,), dtype=float64),
            'robot0_joint_pos_cos': Tensor(shape=(7,), dtype=float64),
            'robot0_joint_pos_sin': Tensor(shape=(7,), dtype=float64),
            'robot0_joint_vel': Tensor(shape=(7,), dtype=float64),
            'robot0_proprio-state': Tensor(shape=(32,), dtype=float64),
            'robot0_robotview_image': Image(shape=(256, 256, 3), dtype=uint8),
        }),
        'reward': float64,
        'tag:placed': bool,
    }),
})
  • Documentazione delle funzionalità :
Caratteristica Classe Forma Tipo D Descrizione
CaratteristicheDict
ID_agente Tensore corda
episodio_id Tensore corda
indice_episodio Tensore int32
passi Set di dati
passi/azione Tensore (7,) galleggiante64
passi/sconto Tensore galleggiante64
passi/immagine Immagine (Nessuno, Nessuno, 3) uint8
passi/è_primo Tensore bool
passi/è_ultimo Tensore bool
passi/è_terminale Tensore bool
passi/osservazione CaratteristicheDict
passi/osservazione/Can_pos Tensore (3,) galleggiante64
passi/osservazione/Can_quat Tensore (4,) galleggiante64
passi/osservazione/Can_to_robot0_eef_pos Tensore (3,) galleggiante64
passi/osservazione/Can_to_robot0_eef_quat Tensore (4,) galleggiante32
passaggi/osservazione/agentview_image Immagine (256, 256, 3) uint8
passi/osservazione/birdview_image Immagine (256, 256, 3) uint8
passi/osservazione/stato-oggetto Tensore (14,) galleggiante64
passi/osservazione/robot0_eef_pos Tensore (3,) galleggiante64
passi/osservazione/robot0_eef_quat Tensore (4,) galleggiante64
passi/osservazione/robot0_eye_in_hand_image Immagine (256, 256, 3) uint8
passi/osservazione/robot0_gripper_qpos Tensore (2,) galleggiante64
passi/osservazione/robot0_gripper_qvel Tensore (2,) galleggiante64
passi/osservazione/robot0_joint_pos_cos Tensore (7,) galleggiante64
passi/osservazione/robot0_joint_pos_sin Tensore (7,) galleggiante64
passi/osservazione/robot0_joint_vel Tensore (7,) galleggiante64
steps/observation/robot0_proprio-state Tensore (32,) galleggiante64
passi/osservazione/robot0_robotview_image Immagine (256, 256, 3) uint8
passi/ricompensa Tensore galleggiante64
passi/tag:posto Tensore bool

robosuite_panda_pick_place_can/synthetic_stochastic_sac_afe13968

  • Descrizione della configurazione : set di dati sintetici generati da un agente stocastico addestrato con SAC (200 episodi).

  • Pagina iniziale : https://github.com/google-research/rlds

  • Dimensione del download : 144.44 MiB

  • Dimensione del set di dati: 622.86 MiB

  • Divisioni :

Diviso Esempi
'train' 200
  • Struttura delle caratteristiche :
FeaturesDict({
    'agent_id': string,
    'episode_id': string,
    'episode_index': int32,
    'steps': Dataset({
        'action': Tensor(shape=(7,), dtype=float32),
        'discount': float64,
        'image': Image(shape=(None, None, 3), dtype=uint8),
        'is_first': bool,
        'is_last': bool,
        'is_terminal': bool,
        'observation': FeaturesDict({
            'Can_pos': Tensor(shape=(3,), dtype=float32),
            'Can_quat': Tensor(shape=(4,), dtype=float32),
            'Can_to_robot0_eef_pos': Tensor(shape=(3,), dtype=float32),
            'Can_to_robot0_eef_quat': Tensor(shape=(4,), dtype=float32),
            'object-state': Tensor(shape=(14,), dtype=float32),
            'robot0_eef_pos': Tensor(shape=(3,), dtype=float32),
            'robot0_eef_quat': Tensor(shape=(4,), dtype=float32),
            'robot0_gripper_qpos': Tensor(shape=(2,), dtype=float32),
            'robot0_gripper_qvel': Tensor(shape=(2,), dtype=float32),
            'robot0_joint_pos_cos': Tensor(shape=(7,), dtype=float32),
            'robot0_joint_pos_sin': Tensor(shape=(7,), dtype=float32),
            'robot0_joint_vel': Tensor(shape=(7,), dtype=float32),
            'robot0_proprio-state': Tensor(shape=(32,), dtype=float32),
        }),
        'reward': float64,
        'tag:placed': bool,
    }),
})
  • Documentazione delle funzionalità :
Caratteristica Classe Forma Tipo D Descrizione
CaratteristicheDict
ID_agente Tensore corda
episodio_id Tensore corda
indice_episodio Tensore int32
passi Set di dati
passi/azione Tensore (7,) galleggiante32
passi/sconto Tensore galleggiante64
passi/immagine Immagine (Nessuno, Nessuno, 3) uint8
passi/è_primo Tensore bool
passi/è_ultimo Tensore bool
passi/è_terminale Tensore bool
passi/osservazione CaratteristicheDict
passi/osservazione/Can_pos Tensore (3,) galleggiante32
passi/osservazione/Can_quat Tensore (4,) galleggiante32
passi/osservazione/Can_to_robot0_eef_pos Tensore (3,) galleggiante32
passi/osservazione/Can_to_robot0_eef_quat Tensore (4,) galleggiante32
passi/osservazione/stato-oggetto Tensore (14,) galleggiante32
passi/osservazione/robot0_eef_pos Tensore (3,) galleggiante32
passi/osservazione/robot0_eef_quat Tensore (4,) galleggiante32
passi/osservazione/robot0_gripper_qpos Tensore (2,) galleggiante32
passi/osservazione/robot0_gripper_qvel Tensore (2,) galleggiante32
passi/osservazione/robot0_joint_pos_cos Tensore (7,) galleggiante32
passi/osservazione/robot0_joint_pos_sin Tensore (7,) galleggiante32
passi/osservazione/robot0_joint_vel Tensore (7,) galleggiante32
steps/observation/robot0_proprio-state Tensore (32,) galleggiante32
passi/ricompensa Tensore galleggiante64
passi/tag:posto Tensore bool