columbia_cairlab_pusht_real

  • Sự miêu tả :

Nhiệm vụ đẩy phẳng UR5

Tách ra Ví dụ
'test' 14
'train' 122
  • Cấu trúc tính năng :
FeaturesDict({
   
'steps': Dataset({
       
'action': FeaturesDict({
           
'gripper_closedness_action': float32,
           
'rotation_delta': Tensor(shape=(3,), dtype=float32, description=Delta change in roll, pitch, yaw.),
           
'terminate_episode': float32,
           
'world_vector': Tensor(shape=(3,), dtype=float32, description=Delta change in XYZ.),
       
}),
       
'is_first': bool,
       
'is_last': bool,
       
'is_terminal': bool,
       
'observation': FeaturesDict({
           
'image': Image(shape=(240, 320, 3), dtype=uint8),
           
'natural_language_embedding': Tensor(shape=(512,), dtype=float32),
           
'natural_language_instruction': string,
           
'robot_state': Tensor(shape=(2,), dtype=float32, description=Robot end effector XY state),
           
'wrist_image': Image(shape=(240, 320, 3), dtype=uint8),
       
}),
       
'reward': Scalar(shape=(), dtype=float32),
   
}),
})
  • Tài liệu tính năng :
Tính năng Lớp học Hình dạng loại D Sự miêu tả
Tính năngDict
bước Tập dữ liệu
bước/hành động Tính năngDict
bước/hành động/gripper_closedness_action Tenxơ phao32 1 nếu kẹp đóng, -1 nếu kẹp mở, 0 nếu không thay đổi.
bước/hành động/xoay_delta Tenxơ (3,) phao32 Thay đổi Delta trong cuộn, cao độ, ngáp.
bước/hành động/chấm dứt_episode Tenxơ phao32
bước/hành động/world_vector Tenxơ (3,) phao32 Thay đổi delta trong XYZ.
bước/là_đầu tiên Tenxơ bool
bước/is_last Tenxơ bool
bước/is_terminal Tenxơ bool
bước/quan sát Tính năngDict
bước/quan sát/hình ảnh Hình ảnh (240, 320, 3) uint8
các bước/quan sát/tự nhiên_ngôn ngữ_embedding Tenxơ (512,) phao32
các bước/quan sát/tự nhiên_ngôn ngữ_hướng dẫn Tenxơ sợi dây
bước/quan sát/robot_state Tenxơ (2,) phao32 Trạng thái tác động cuối của robot XY
bước/quan sát/hình ảnh cổ tay Hình ảnh (240, 320, 3) uint8
bước/phần thưởng vô hướng phao32
  • Trích dẫn :
@inproceedings{chi2023diffusionpolicy,
    title
={Diffusion Policy: Visuomotor Policy Learning via Action Diffusion},
    author
={Chi, Cheng and Feng, Siyuan and Du, Yilun and Xu, Zhenjia and Cousineau, Eric and Burchfiel, Benjamin and Song, Shuran},
    booktitle
={Proceedings of Robotics: Science and Systems (RSS)},
    year
={2023}
}