фрактал20220817_данные

  • Описание :

Настольная манипуляция с 17 объектами

Расколоть Примеры
'train' 87 212
  • Структура функции :
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, description=continuous gripper position),
           
'rotation_delta': Tensor(shape=(3,), dtype=float32, description=rpy commanded orientation displacement, in base-relative frame),
           
'terminate_episode': Tensor(shape=(3,), dtype=int32),
           
'world_vector': Tensor(shape=(3,), dtype=float32, description=commanded end-effector displacement, in base-relative frame),
       
}),
       
'is_first': bool,
       
'is_last': bool,
       
'is_terminal': bool,
       
'observation': FeaturesDict({
           
'base_pose_tool_reached': Tensor(shape=(7,), dtype=float32, description=end-effector base-relative position+quaternion pose),
           
'gripper_closed': Tensor(shape=(1,), dtype=float32),
           
'gripper_closedness_commanded': Tensor(shape=(1,), dtype=float32, description=continuous gripper position),
           
'height_to_bottom': Tensor(shape=(1,), dtype=float32, description=height of end-effector from ground),
           
'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, description=rotational displacement from current orientation to target),
           
'src_rotation': Tensor(shape=(4,), dtype=float32),
           
'vector_to_go': Tensor(shape=(3,), dtype=float32, description=displacement from current end-effector position to target),
           
'workspace_bounds': Tensor(shape=(3, 3), dtype=float32),
       
}),
       
'reward': Scalar(shape=(), dtype=float32),
   
}),
})
  • Функциональная документация :
Особенность Сорт Форма Дтип Описание
ВозможностиDict
аспекты ВозможностиDict Аспекты сессии для рейтингов краудкомпьютеров
аспекты/уже_успех Тензор логическое значение
аспекты/выполнимость Тензор логическое значение
аспекты/has_aspects Тензор логическое значение
аспекты/успех Тензор логическое значение
аспекты/нежелательные Тензор логическое значение
атрибуты ВозможностиDict
атрибуты/режим_коллекции Тензор int64
атрибуты/имя_режима_коллекции Тензор нить
атрибуты/тип_данных Тензор int64
атрибуты/имя_типа_данных Тензор нить
атрибуты/окр. Тензор int64
атрибуты/имя_окружения Тензор нить
атрибуты/местоположение Тензор int64
атрибуты/имя_локации Тензор нить
атрибуты/семейство_объектов Тензор int64
атрибуты/имя_семейства_объектов Тензор нить
атрибуты/семейство_задач Тензор int64
атрибуты/имя_семейства_задач Тензор нить
шаги Набор данных
шаги/действия ВозможностиDict
шаги/действие/base_displacement_vector Тензор (2,) float32
шаги/действие/base_displacement_vertical_rotation Тензор (1,) float32
шаги/действие/gripper_closedness_action Тензор (1,) float32 постоянное положение захвата
шаги/действие/rotation_delta Тензор (3,) float32 rpy командовал смещением ориентации в базовой системе координат
шаги/действие/terminate_episode Тензор (3,) int32
шаги/действие/world_vector Тензор (3,) float32 управляемое смещение рабочего органа в кадре относительно основания
шаги/is_first Тензор логическое значение
шаги/is_last Тензор логическое значение
шаги/is_terminal Тензор логическое значение
шаги/наблюдение ВозможностиDict
шаги/наблюдение/base_pose_tool_reached Тензор (7,) float32 положение относительно основания конечного эффектора + поза кватерниона
шаги/наблюдение/gripper_closed Тензор (1,) float32
шаги/наблюдение/gripper_closedness_commanded Тензор (1,) float32 постоянное положение захвата
шаги/наблюдение/height_to_bottom Тензор (1,) float32 высота рабочего органа от земли
шаги/наблюдение/изображение Изображение (256, 320, 3) uint8
шаги/наблюдение/natural_language_embedding Тензор (512,) float32
шаги/наблюдение/natural_language_instruction Тензор нить
шаги/наблюдение/orientation_box Тензор (2, 3) float32
шаги/наблюдение/orientation_start Тензор (4,) float32
шаги/наблюдение/robot_orientation_positions_box Тензор (3, 3) float32
шаги/наблюдение/rotation_delta_to_go Тензор (3,) float32 вращательное смещение от текущей ориентации к цели
шаги/наблюдение/src_rotation Тензор (4,) float32
шаги/наблюдение/vector_to_go Тензор (3,) float32 смещение от текущего положения конечного эффектора к цели
шаги/наблюдение/workspace_bounds Тензор (3, 3) float32
шаги/награда Скаляр float32
  • Цитата :
@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}
}