| --- |
| size_categories: |
| - 10K<n<100K |
| tags: |
| - Robotics |
| - Pre-training |
| --- |
| # Fractal20220817 Dataset |
|
|
| The **[Fractal20220817](https://research.google/blog/rt-1-robotics-transformer-for-real-world-control-at-scale/)** dataset is provided in **RLDS format** and contains monocular depth annotations generated using **[Video-Depth-Anything](https://github.com/DepthAnything/Video-Depth-Anything)**(Model: Video-Depth-Anything-V2-Small). |
|
|
| - **Format**: RLDS |
| - **Depth Annotation method**: Video-Depth-Anything (monocular depth estimation) |
| - **Total size**: 154 GB |
|
|
| feature.json |
| ```json |
| { |
| "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict", |
| "featuresDict": { |
| "features": { |
| "attributes": { |
| "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict", |
| "featuresDict": { |
| "features": { |
| "collection_mode_name": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "string", |
| "encoding": "none" |
| } |
| }, |
| "collection_mode": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "int64", |
| "encoding": "none" |
| } |
| }, |
| "task_family_name": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "string", |
| "encoding": "none" |
| } |
| }, |
| "location_name": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "string", |
| "encoding": "none" |
| } |
| }, |
| "env_name": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "string", |
| "encoding": "none" |
| } |
| }, |
| "env": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "int64", |
| "encoding": "none" |
| } |
| }, |
| "objects_family_name": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "string", |
| "encoding": "none" |
| } |
| }, |
| "data_type": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "int64", |
| "encoding": "none" |
| } |
| }, |
| "objects_family": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "int64", |
| "encoding": "none" |
| } |
| }, |
| "location": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "int64", |
| "encoding": "none" |
| } |
| }, |
| "task_family": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "int64", |
| "encoding": "none" |
| } |
| }, |
| "data_type_name": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "string", |
| "encoding": "none" |
| } |
| } |
| } |
| } |
| }, |
| "aspects": { |
| "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict", |
| "featuresDict": { |
| "features": { |
| "feasible": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| }, |
| "success": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| }, |
| "undesirable": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| }, |
| "already_success": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| }, |
| "has_aspects": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| } |
| } |
| }, |
| "description": "Session Aspects for crowdcompute ratings" |
| }, |
| "steps": { |
| "pythonClassName": "tensorflow_datasets.core.features.dataset_feature.Dataset", |
| "sequence": { |
| "feature": { |
| "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict", |
| "featuresDict": { |
| "features": { |
| "is_terminal": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| }, |
| "is_first": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| }, |
| "action": { |
| "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict", |
| "featuresDict": { |
| "features": { |
| "gripper_closedness_action": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "1" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "continuous gripper position" |
| }, |
| "terminate_episode": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "3" |
| ] |
| }, |
| "dtype": "int32", |
| "encoding": "none" |
| } |
| }, |
| "base_displacement_vector": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "2" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "rotation_delta": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "3" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "rpy commanded orientation displacement, in base-relative frame" |
| }, |
| "base_displacement_vertical_rotation": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "1" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "world_vector": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "3" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "commanded end-effector displacement, in base-relative frame" |
| } |
| } |
| } |
| }, |
| "reward": { |
| "pythonClassName": "tensorflow_datasets.core.features.scalar.Scalar", |
| "tensor": { |
| "shape": {}, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "observation": { |
| "pythonClassName": "tensorflow_datasets.core.features.features_dict.FeaturesDict", |
| "featuresDict": { |
| "features": { |
| "rotation_delta_to_go": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "3" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "rotational displacement from current orientation to target" |
| }, |
| "orientation_start": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "4" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "height_to_bottom": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "1" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "height of end-effector from ground" |
| }, |
| "natural_language_instruction": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "string", |
| "encoding": "none" |
| } |
| }, |
| "natural_language_embedding": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "512" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "workspace_bounds": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "3", |
| "3" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "robot_orientation_positions_box": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "3", |
| "3" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "gripper_closedness_commanded": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "1" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "continuous gripper position" |
| }, |
| "base_pose_tool_reached": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "7" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "end-effector base-relative position+quaternion pose" |
| }, |
| "vector_to_go": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "3" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| }, |
| "description": "displacement from current end-effector position to target" |
| }, |
| "src_rotation": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "4" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "gripper_closed": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "1" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| }, |
| "image": { |
| "pythonClassName": "tensorflow_datasets.core.features.image_feature.Image", |
| "image": { |
| "shape": { |
| "dimensions": [ |
| "256", |
| "320", |
| "3" |
| ] |
| }, |
| "dtype": "uint8", |
| "encodingFormat": "jpeg" |
| } |
| }, |
| "image_depth": { |
| "pythonClassName": "tensorflow_datasets.core.features.image_feature.Image", |
| "image": { |
| "shape": { |
| "dimensions": [ |
| "256", |
| "256", |
| "3" |
| ] |
| }, |
| "dtype": "uint8", |
| "encodingFormat": "jpeg" |
| } |
| }, |
| "orientation_box": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": { |
| "dimensions": [ |
| "2", |
| "3" |
| ] |
| }, |
| "dtype": "float32", |
| "encoding": "none" |
| } |
| } |
| } |
| } |
| }, |
| "is_last": { |
| "pythonClassName": "tensorflow_datasets.core.features.tensor_feature.Tensor", |
| "tensor": { |
| "shape": {}, |
| "dtype": "bool", |
| "encoding": "none" |
| } |
| } |
| } |
| } |
| }, |
| "length": "-1" |
| } |
| } |
| } |
| } |
| } |
| ``` |
