---
license: apache-2.0
task_categories:
- robotics
tags:
- LeRobot
- robomme
- lerobot
- geometry
- camera-calibration
configs:
- config_name: default
data_files: data/*/*.parquet
---
This dataset was created using [LeRobot](https://github.com/huggingface/lerobot).
## Dataset Description
# RoboMME geometry augmentation
Generated from `lerobot/robomme` at revision `main`.
The original RGB, state, action, annotation, episode and task columns are
unchanged.
Geometry is recovered from the original RoboMME HDF5 trajectories. Every HDF5
episode is matched to its LeRobot episode using recorded joint state/actions,
then state, action, and demo status are checked frame by frame before geometry
is copied.
## Added end-effector columns
Pose columns contain:
[position, rotation, gripper]
The following rotation representations are available:
- rotvec
- quat_xyzw
- rot6d
- rot6d_rows
- matrix
- euler_xyz_extrinsic
`observation.state.ee_*` contains the recorded absolute measured end-effector
pose.
`action.ee_absolute.*` contains RoboMME's recorded absolute end-effector target
corresponding to the joint-space action. RoboMME computed and stored this pose
from `joint_action` using forward kinematics during dataset recording; this augmentation script does not recompute kinematics.
## Cameras
Added columns:
- `observation.intrinsics.image`
- `observation.intrinsics.wrist_image`
- `observation.extrinsics.image`
- `observation.extrinsics.wrist_image`
Intrinsics are 3x3 OpenCV camera matrices.
Extrinsics are stored as world-from-camera `[xyz, quaternion xyzw]`. RoboMME's
recorded OpenCV `extrinsic_cv` matrices are inverted to obtain this convention.
## Gripper events
- `observation.gripper.time_to_close`
- `observation.gripper.time_to_open`
Values are seconds until the next gripper command transition.
RoboMME's action convention is:
- -1 = close
- +1 = open
Event search never crosses an episode boundary or the conditioning-demo /
execution boundary. NaN means no subsequent event in the current segment.
## Chunk-anchored statistics
`meta/stats.json` additionally contains per-offset statistics under:
action.ee_absolute..anchored_horizon_50
Only execution chunks are used; conditioning-video frames are excluded.
Positions are anchored to the measured end-effector position at the start of
the chunk:
p_relative[t] = p_action[t] - p_state[start]
Rotations are:
R_relative[t] = R_action[t] @ R_state[start].T
The gripper remains an absolute command.
These are stats keys, not dataset feature columns.
- **Homepage:** [More Information Needed]
- **Paper:** [More Information Needed]
- **License:** apache-2.0
## Dataset Structure
[meta/info.json](meta/info.json):
```json
{
"codebase_version": "v3.0",
"fps": 10,
"features": {
"image": {
"dtype": "video",
"shape": [
256,
256,
3
],
"names": [
"height",
"width",
"channel"
],
"fps": 10,
"info": {
"video.height": 256,
"video.width": 256,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"video.fps": 10,
"video.channels": 3,
"has_audio": false
}
},
"wrist_image": {
"dtype": "video",
"shape": [
256,
256,
3
],
"names": [
"height",
"width",
"channel"
],
"fps": 10,
"info": {
"video.height": 256,
"video.width": 256,
"video.codec": "av1",
"video.pix_fmt": "yuv420p",
"video.is_depth_map": false,
"video.fps": 10,
"video.channels": 3,
"has_audio": false
}
},
"state": {
"dtype": "float32",
"shape": [
8
],
"names": [
"state"
],
"fps": 10
},
"actions": {
"dtype": "float32",
"shape": [
8
],
"names": [
"actions"
],
"fps": 10
},
"exec_start_idx": {
"dtype": "int32",
"shape": [
1
],
"names": [
"execution start idx in current episode"
],
"fps": 10
},
"is_demo": {
"dtype": "bool",
"shape": [
1
],
"names": [
"is video demo"
],
"fps": 10
},
"step_idx": {
"dtype": "int32",
"shape": [
1
],
"names": [
"step_idx"
],
"fps": 10
},
"epis_idx": {
"dtype": "int32",
"shape": [
1
],
"names": [
"epis_idx"
],
"fps": 10
},
"simple_subgoal": {
"dtype": "string",
"shape": [
1
],
"names": [
"simple subgoal"
],
"fps": 10
},
"grounded_subgoal": {
"dtype": "string",
"shape": [
1
],
"names": [
"detailed subgoal with grounding information, e.g, object location"
],
"fps": 10
},
"simple_subgoal_online": {
"dtype": "string",
"shape": [
1
],
"names": [
"simple subgoal online"
],
"fps": 10
},
"grounded_subgoal_online": {
"dtype": "string",
"shape": [
1
],
"names": [
"detailed subgoal with grounding information, e.g, object location"
],
"fps": 10
},
"timestamp": {
"dtype": "float32",
"shape": [
1
],
"names": null
},
"frame_index": {
"dtype": "int64",
"shape": [
1
],
"names": null
},
"episode_index": {
"dtype": "int64",
"shape": [
1
],
"names": null
},
"index": {
"dtype": "int64",
"shape": [
1
],
"names": null
},
"task_index": {
"dtype": "int64",
"shape": [
1
],
"names": null
},
"observation.state.ee_rotvec": {
"dtype": "float32",
"shape": [
7
],
"names": null
},
"action.ee_absolute.rotvec": {
"dtype": "float32",
"shape": [
7
],
"names": null
},
"observation.state.ee_quat_xyzw": {
"dtype": "float32",
"shape": [
8
],
"names": null
},
"action.ee_absolute.quat_xyzw": {
"dtype": "float32",
"shape": [
8
],
"names": null
},
"observation.state.ee_rot6d": {
"dtype": "float32",
"shape": [
10
],
"names": null
},
"action.ee_absolute.rot6d": {
"dtype": "float32",
"shape": [
10
],
"names": null
},
"observation.state.ee_rot6d_rows": {
"dtype": "float32",
"shape": [
10
],
"names": null
},
"action.ee_absolute.rot6d_rows": {
"dtype": "float32",
"shape": [
10
],
"names": null
},
"observation.state.ee_matrix": {
"dtype": "float32",
"shape": [
13
],
"names": null
},
"action.ee_absolute.matrix": {
"dtype": "float32",
"shape": [
13
],
"names": null
},
"observation.state.ee_euler_xyz_extrinsic": {
"dtype": "float32",
"shape": [
7
],
"names": null
},
"action.ee_absolute.euler_xyz_extrinsic": {
"dtype": "float32",
"shape": [
7
],
"names": null
},
"observation.gripper.time_to_close": {
"dtype": "float32",
"shape": [
1
],
"names": null
},
"observation.gripper.time_to_open": {
"dtype": "float32",
"shape": [
1
],
"names": null
},
"observation.intrinsics.image": {
"dtype": "float32",
"shape": [
3,
3
],
"names": null
},
"observation.intrinsics.wrist_image": {
"dtype": "float32",
"shape": [
3,
3
],
"names": null
},
"observation.extrinsics.image": {
"dtype": "float32",
"shape": [
7
],
"names": null
},
"observation.extrinsics.wrist_image": {
"dtype": "float32",
"shape": [
7
],
"names": null
}
},
"total_episodes": 1600,
"total_frames": 768897,
"total_tasks": 116,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 200,
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
"robot_type": "panda",
"splits": {
"train": "0:1600"
}
}
```
## Citation
**BibTeX:**
```bibtex
[More Information Needed]
```