<br />

- **Description**:

Teleoped Google robot doing mostly pick-place from a table

- **Homepage** :
  <https://www.kaggle.com/datasets/google/bc-z-robot/discussion/309201>

- **Source code** :
  [`tfds.robotics.rtx.BcZ`](https://github.com/tensorflow/datasets/tree/master/tensorflow_datasets/robotics/rtx/rtx.py)

- **Versions**:

  - **`0.1.0`** (default): Initial release.
- **Download size** : `Unknown size`

- **Dataset size** : `81.15 GiB`

- **Auto-cached**
  ([documentation](https://www.tensorflow.org/datasets/performances#auto-caching)):
  No

- **Splits**:

| Split | Examples |
|---|---|
| `'train'` | 39,350 |
| `'val'` | 3,914 |

- **Feature structure**:

    FeaturesDict({
        'steps': Dataset({
            'action': FeaturesDict({
                'future/axis_angle_residual': Tensor(shape=(30,), dtype=float32, description=The next 10 actions for the rotation. Each action is a 3D delta to add to the current axis angle.),
                'future/target_close': Tensor(shape=(10,), dtype=int64, description=The next 10 actions for the gripper. Each action is the value the gripper closure should be changed to (notably it is *not* a delta.)),
                'future/xyz_residual': Tensor(shape=(30,), dtype=float32, description=The next 10 actions for the positions. Each action is a 3D delta to add to current position.),
            }),
            'is_first': bool,
            'is_last': bool,
            'is_terminal': bool,
            'observation': FeaturesDict({
                'episode_success': float32,
                'image': Image(shape=(171, 213, 3), dtype=uint8, description=Camera image of the robot, downsampled 3x),
                'natural_language_embedding': Tensor(shape=(512,), dtype=float32, description=An embedding of the task via Universal Sentence Encoder (https://tfhub.dev/google/universal-sentence-encoder/4)),
                'natural_language_instruction': string,
                'present/autonomous': int64,
                'present/axis_angle': Tensor(shape=(3,), dtype=float32, description=The current rotation of the end effector in axis-angle representation.),
                'present/intervention': int64,
                'present/sensed_close': Tensor(shape=(1,), dtype=float32, description=How much the gripper is currently closed. Scaled from 0 to 1, but not all values from 0 to 1 are reachable. The range in the data is about 0.2 to 1),
                'present/xyz': Tensor(shape=(3,), dtype=float32, description=The current position of the end effector in axis-angle representation, in robot frame),
                'sequence_length': int64,
            }),
            'reward': Scalar(shape=(), dtype=float32),
        }),
    })

- **Feature documentation**:

| Feature | Class | Shape | Dtype | Description |
|---|---|---|---|---|
|   | FeaturesDict |   |   |   |
| steps | Dataset |   |   |   |
| steps/action | FeaturesDict |   |   |   |
| steps/action/future/axis_angle_residual | Tensor | (30,) | float32 | The next 10 actions for the rotation. Each action is a 3D delta to add to the current axis angle. |
| steps/action/future/target_close | Tensor | (10,) | int64 | The next 10 actions for the gripper. Each action is the value the gripper closure should be changed to (notably it is *not* a delta.) |
| steps/action/future/xyz_residual | Tensor | (30,) | float32 | The next 10 actions for the positions. Each action is a 3D delta to add to current position. |
| steps/is_first | Tensor |   | bool |   |
| steps/is_last | Tensor |   | bool |   |
| steps/is_terminal | Tensor |   | bool |   |
| steps/observation | FeaturesDict |   |   |   |
| steps/observation/episode_success | Tensor |   | float32 | A 0-1 success label |
| steps/observation/image | Image | (171, 213, 3) | uint8 | Camera image of the robot, downsampled 3x |
| steps/observation/natural_language_embedding | Tensor | (512,) | float32 | An embedding of the task via Universal Sentence Encoder (https://tfhub.dev/google/universal-sentence-encoder/4) |
| steps/observation/natural_language_instruction | Tensor |   | string | The task the robot was asked to do. |
| steps/observation/present/autonomous | Tensor |   | int64 | Episodes are collected via DAgger. This is a 0/1 label for whether the action is from the policy or the teleoperator. 1 = from policy. |
| steps/observation/present/axis_angle | Tensor | (3,) | float32 | The current rotation of the end effector in axis-angle representation. |
| steps/observation/present/intervention | Tensor |   | int64 | Episodes are collected via DAgger. This is a 0/1 label for whether the action is from the policy or the teleoperator. 1 = from teleoperator. This is exactly the opposite of present/autonomous |
| steps/observation/present/sensed_close | Tensor | (1,) | float32 | How much the gripper is currently closed. Scaled from 0 to 1, but not all values from 0 to 1 are reachable. The range in the data is about 0.2 to 1 |
| steps/observation/present/xyz | Tensor | (3,) | float32 | The current position of the end effector in axis-angle representation, in robot frame |
| steps/observation/sequence_length | Tensor |   | int64 | Length of the episode |
| steps/reward | Scalar |   | float32 |   |

- **Supervised keys** (See
  [`as_supervised` doc](https://www.tensorflow.org/datasets/api_docs/python/tfds/load#args)):
  `None`

- **Figure**
  ([tfds.show_examples](https://www.tensorflow.org/datasets/api_docs/python/tfds/visualization/show_examples)):
  Not supported.

- **Examples**
  ([tfds.as_dataframe](https://www.tensorflow.org/datasets/api_docs/python/tfds/as_dataframe)):

<iframe src="https:///frame/datasets/catalog/bc_z_68b58f01b6ab13d7f782173c9d30a267becfd6b1d6cce1ebb49534e1eb1f94a9.frame" class="framebox inherit-locale " allow="clipboard-write https://" allowfullscreen is-upgraded></iframe>

- **Citation**:

    @inproceedings{jang2021bc,
    title={ {BC}-Z: Zero-Shot Task Generalization with Robotic Imitation Learning},
    author={Eric Jang and Alex Irpan and Mohi Khansari and Daniel Kappler and Frederik Ebert and Corey Lynch and Sergey Levine and Chelsea Finn},
    booktitle={5th Annual Conference on Robot Learning},
    year={2021},
    url={https://openreview.net/forum?id=8kbp23tSGYv} }