Upload folder using huggingface_hub
Browse files- example_data/umi_data_sample/README.md +102 -0
- example_data/umi_data_sample/README_EN.md +109 -0
- example_data/umi_data_sample/annotation/action_steps_000001.jsonl +4 -4
- example_data/umi_data_sample/annotation/action_steps_000002.jsonl +18 -18
- example_data/umi_data_sample/annotation/action_steps_000003.jsonl +8 -8
- example_data/umi_data_sample/annotation/action_steps_000004.jsonl +27 -27
- example_data/umi_data_sample/data/chunk-000/episode_000000.parquet +2 -2
- example_data/umi_data_sample/data/chunk-000/episode_000001.parquet +2 -2
- example_data/umi_data_sample/data/chunk-000/episode_000002.parquet +2 -2
- example_data/umi_data_sample/data/chunk-000/episode_000003.parquet +2 -2
- example_data/umi_data_sample/data/chunk-000/episode_000004.parquet +2 -2
- example_data/umi_data_sample/meta/calibration.json +4 -4
example_data/umi_data_sample/README.md
ADDED
|
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-sa-4.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# 启元机器人家庭操作挑战赛数据集
|
| 6 |
+
|
| 7 |
+
> English version: [README_EN.md](README_EN.md)
|
| 8 |
+
|
| 9 |
+
## UMI Sample Data
|
| 10 |
+
|
| 11 |
+
本目录是一个标准的 **LeRobot v2.1** 数据集，包含 5 个双手操作 episode，以及左、右手腕载 ego 相机数据。
|
| 12 |
+
|
| 13 |
+
```bash
|
| 14 |
+
pip install "lerobot==0.3.3"
|
| 15 |
+
```
|
| 16 |
+
|
| 17 |
+
```python
|
| 18 |
+
from lerobot.datasets.lerobot_dataset import LeRobotDataset
|
| 19 |
+
|
| 20 |
+
dataset = LeRobotDataset(
|
| 21 |
+
repo_id="local/umi_sample_data_v21",
|
| 22 |
+
root="/path/to/umi_sample_data_v21",
|
| 23 |
+
)
|
| 24 |
+
```
|
| 25 |
+
|
| 26 |
+
### 数据概览
|
| 27 |
+
|
| 28 |
+
| Episode | `task_index` | 任务 | 帧数 | 时长 |
|
| 29 |
+
| :--- | ---: | :--- | ---: | ---: |
|
| 30 |
+
| `episode_000000` | 0 | fold the red shirt | 1410 | 47 s |
|
| 31 |
+
| `episode_000001` | 1 | fold the black shirt | 1050 | 35 s |
|
| 32 |
+
| `episode_000002` | 2 | fold the yellow shirt | 870 | 29 s |
|
| 33 |
+
| `episode_000003` | 1 | fold the black shirt | 1050 | 35 s |
|
| 34 |
+
| `episode_000004` | 3 | fold the brown shirt | 1560 | 52 s |
|
| 35 |
+
|
| 36 |
+
数据集共 5940 帧、4 个唯一任务，帧率为 30 FPS，ego 视频分辨率为 960 × 960。
|
| 37 |
+
|
| 38 |
+
### 目录结构
|
| 39 |
+
|
| 40 |
+
```text
|
| 41 |
+
umi_sample_data_v21/
|
| 42 |
+
├── data/chunk-000/ # 5 个 episode Parquet 文件
|
| 43 |
+
├── videos/chunk-000/
|
| 44 |
+
│ ├── observation.images.left_ego/ # 左手 ego 视频
|
| 45 |
+
│ └── observation.images.right_ego/ # 右手 ego 视频
|
| 46 |
+
├── meta/
|
| 47 |
+
│ ├── info.json # 数据集与字段定义
|
| 48 |
+
│ ├── tasks.jsonl # 任务与 task_index 映射
|
| 49 |
+
│ ├── episodes.jsonl # episode 长度与任务
|
| 50 |
+
│ ├── episodes_stats.jsonl # 每个 episode 的统计量
|
| 51 |
+
│ └── calibration.json # 相机与 IMU 标定参数
|
| 52 |
+
├── annotation/ # episode 级任务和动作分段标注
|
| 53 |
+
└── imu/ # 左右手 IMU 数据
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
### 数据集字段说明
|
| 57 |
+
|
| 58 |
+
#### 图像
|
| 59 |
+
|
| 60 |
+
| 数据集字段 | 数据源 |
|
| 61 |
+
| :--- | :--- |
|
| 62 |
+
| `observation.images.left_ego` | 左手腕载 RGB ego 相机，960 × 960 |
|
| 63 |
+
| `observation.images.right_ego` | 右手腕载 RGB ego 相机，960 × 960 |
|
| 64 |
+
|
| 65 |
+
#### 本体感知与动作
|
| 66 |
+
|
| 67 |
+
`observation.state` 和 `action` 均为 16 维，字段顺序一致：
|
| 68 |
+
|
| 69 |
+
| 索引 | 字段顺序 | 物理意义 | 单位 |
|
| 70 |
+
| :--- | :--- | :--- | :--- |
|
| 71 |
+
| 0–2 | `left_x, left_y, left_z` | 左手末端位置 | m |
|
| 72 |
+
| 3–6 | `left_qw, left_qx, left_qy, left_qz` | 左手末端四元数 `(w, x, y, z)` | - |
|
| 73 |
+
| 7 | `left_gripper` | 左夹爪开合角度 | ° |
|
| 74 |
+
| 8–10 | `right_x, right_y, right_z` | 右手末端位置 | m |
|
| 75 |
+
| 11–14 | `right_qw, right_qx, right_qy, right_qz` | 右手末端四元数 `(w, x, y, z)` | - |
|
| 76 |
+
| 15 | `right_gripper` | 右夹爪开合角度 | ° |
|
| 77 |
+
|
| 78 |
+
`observation.state` 表示当前帧状态。除末帧外，`action[t] = state[t+1]`；末帧 action 保留原始采集序列的下一时刻目标，因此不一定等于本 episode 的末帧 state。左右手位姿使用独立坐标系，不能直接计算双手之间的相对距离或姿态。
|
| 79 |
+
|
| 80 |
+
#### 索引字段
|
| 81 |
+
|
| 82 |
+
| 字段 | 类型 | 说明 |
|
| 83 |
+
| :--- | :--- | :--- |
|
| 84 |
+
| `timestamp` | float32 | episode 内时间，单位为秒 |
|
| 85 |
+
| `frame_index` | int64 | episode 内帧编号，从 0 开始 |
|
| 86 |
+
| `episode_index` | int64 | episode 编号，范围为 0–4 |
|
| 87 |
+
| `index` | int64 | 数据集全局帧编号，范围为 0–5939 |
|
| 88 |
+
| `task_index` | int64 | 任务编号，对应 `meta/tasks.jsonl` |
|
| 89 |
+
|
| 90 |
+
#### 标注与 IMU
|
| 91 |
+
|
| 92 |
+
| 文件 | 说明 |
|
| 93 |
+
| :--- | :--- |
|
| 94 |
+
| `annotation/episode_subtasks_*.jsonl` | episode 级任务、目标物体与成功状态 |
|
| 95 |
+
| `annotation/action_steps_*.jsonl` | 细粒度动作分段 |
|
| 96 |
+
| `imu/episode_*_{left,right}.csv` | 左右手时间戳、三轴角速度和三轴加速度 |
|
| 97 |
+
|
| 98 |
+
标注区间采用 `[start_frame_index, end_frame_index)`，即包含起始帧、不包含结束帧。
|
| 99 |
+
|
| 100 |
+
`action_steps` 的 `subtask_index` 在本批交付里**恒为 0**，是占位字段，不要当作指向
|
| 101 |
+
`episode_subtasks` 的外键使用；两层的对应关系请按帧区间重叠自行计算。本数据集每集
|
| 102 |
+
恰好 1 条 `episode_subtasks`（=整集任务）。
|
example_data/umi_data_sample/README_EN.md
ADDED
|
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-sa-4.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# Qiyuan Robotics Home Manipulation Challenge Dataset
|
| 6 |
+
|
| 7 |
+
> 中文版本：[README.md](README.md)
|
| 8 |
+
|
| 9 |
+
## UMI Sample Data
|
| 10 |
+
|
| 11 |
+
This directory is a standard **LeRobot v2.1** dataset containing 5 bimanual manipulation
|
| 12 |
+
episodes, together with left and right wrist-mounted ego-camera data.
|
| 13 |
+
|
| 14 |
+
```bash
|
| 15 |
+
pip install "lerobot==0.3.3"
|
| 16 |
+
```
|
| 17 |
+
|
| 18 |
+
```python
|
| 19 |
+
from lerobot.datasets.lerobot_dataset import LeRobotDataset
|
| 20 |
+
|
| 21 |
+
dataset = LeRobotDataset(
|
| 22 |
+
repo_id="local/umi_sample_data_v21",
|
| 23 |
+
root="/path/to/umi_sample_data_v21",
|
| 24 |
+
)
|
| 25 |
+
```
|
| 26 |
+
|
| 27 |
+
### Overview
|
| 28 |
+
|
| 29 |
+
| Episode | `task_index` | Task | Frames | Duration |
|
| 30 |
+
| :--- | ---: | :--- | ---: | ---: |
|
| 31 |
+
| `episode_000000` | 0 | fold the red shirt | 1410 | 47 s |
|
| 32 |
+
| `episode_000001` | 1 | fold the black shirt | 1050 | 35 s |
|
| 33 |
+
| `episode_000002` | 2 | fold the yellow shirt | 870 | 29 s |
|
| 34 |
+
| `episode_000003` | 1 | fold the black shirt | 1050 | 35 s |
|
| 35 |
+
| `episode_000004` | 3 | fold the brown shirt | 1560 | 52 s |
|
| 36 |
+
|
| 37 |
+
The dataset has 5940 frames and 4 unique tasks, at 30 FPS, with 960 × 960 ego video.
|
| 38 |
+
|
| 39 |
+
### Directory structure
|
| 40 |
+
|
| 41 |
+
```text
|
| 42 |
+
umi_sample_data_v21/
|
| 43 |
+
├── data/chunk-000/ # 5 episode Parquet files
|
| 44 |
+
├── videos/chunk-000/
|
| 45 |
+
│ ├── observation.images.left_ego/ # left-hand ego video
|
| 46 |
+
│ └── observation.images.right_ego/ # right-hand ego video
|
| 47 |
+
├── meta/
|
| 48 |
+
│ ├── info.json # dataset and field definitions
|
| 49 |
+
│ ├── tasks.jsonl # task to task_index mapping
|
| 50 |
+
│ ├── episodes.jsonl # episode lengths and tasks
|
| 51 |
+
│ ├── episodes_stats.jsonl # per-episode statistics
|
| 52 |
+
│ └── calibration.json # camera and IMU calibration
|
| 53 |
+
├── annotation/ # episode-level task and action-segment annotation
|
| 54 |
+
└── imu/ # left and right IMU data
|
| 55 |
+
```
|
| 56 |
+
|
| 57 |
+
### Field reference
|
| 58 |
+
|
| 59 |
+
#### Images
|
| 60 |
+
|
| 61 |
+
| Dataset field | Source |
|
| 62 |
+
| :--- | :--- |
|
| 63 |
+
| `observation.images.left_ego` | Left wrist-mounted RGB ego camera, 960 × 960 |
|
| 64 |
+
| `observation.images.right_ego` | Right wrist-mounted RGB ego camera, 960 × 960 |
|
| 65 |
+
|
| 66 |
+
#### Proprioception and actions
|
| 67 |
+
|
| 68 |
+
`observation.state` and `action` are both 16-dimensional, in the same field order:
|
| 69 |
+
|
| 70 |
+
| Indices | Field order | Meaning | Unit |
|
| 71 |
+
| :--- | :--- | :--- | :--- |
|
| 72 |
+
| 0–2 | `left_x, left_y, left_z` | Left end-effector position | m |
|
| 73 |
+
| 3–6 | `left_qw, left_qx, left_qy, left_qz` | Left end-effector quaternion `(w, x, y, z)` | - |
|
| 74 |
+
| 7 | `left_gripper` | Left gripper opening angle | ° |
|
| 75 |
+
| 8–10 | `right_x, right_y, right_z` | Right end-effector position | m |
|
| 76 |
+
| 11–14 | `right_qw, right_qx, right_qy, right_qz` | Right end-effector quaternion `(w, x, y, z)` | - |
|
| 77 |
+
| 15 | `right_gripper` | Right gripper opening angle | ° |
|
| 78 |
+
|
| 79 |
+
`observation.state` is the current-frame state. Except for the final frame,
|
| 80 |
+
`action[t] = state[t+1]`; the final frame's action keeps the next target from the original
|
| 81 |
+
capture sequence, so it is not necessarily equal to this episode's final state. Left and right
|
| 82 |
+
poses live in independent coordinate frames — you cannot directly compute the relative distance
|
| 83 |
+
or orientation between the two hands.
|
| 84 |
+
|
| 85 |
+
#### Index fields
|
| 86 |
+
|
| 87 |
+
| Field | Type | Description |
|
| 88 |
+
| :--- | :--- | :--- |
|
| 89 |
+
| `timestamp` | float32 | Time within the episode, in seconds |
|
| 90 |
+
| `frame_index` | int64 | Frame number within the episode, 0-based |
|
| 91 |
+
| `episode_index` | int64 | Episode number, range 0–4 |
|
| 92 |
+
| `index` | int64 | Global frame number in the dataset, range 0–5939 |
|
| 93 |
+
| `task_index` | int64 | Task number, keyed to `meta/tasks.jsonl` |
|
| 94 |
+
|
| 95 |
+
#### Annotations and IMU
|
| 96 |
+
|
| 97 |
+
| File | Description |
|
| 98 |
+
| :--- | :--- |
|
| 99 |
+
| `annotation/episode_subtasks_*.jsonl` | Episode-level task, target objects and success flag |
|
| 100 |
+
| `annotation/action_steps_*.jsonl` | Fine-grained action segments |
|
| 101 |
+
| `imu/episode_*_{left,right}.csv` | Left and right timestamps, three-axis angular velocity and acceleration |
|
| 102 |
+
|
| 103 |
+
Annotation intervals use the half-open convention `[start_frame_index, end_frame_index)` —
|
| 104 |
+
the start frame is included, the end frame is not.
|
| 105 |
+
|
| 106 |
+
The `subtask_index` field of `action_steps` is **constantly 0** across this delivery. It is a
|
| 107 |
+
placeholder — do not use it as a foreign key into `episode_subtasks`; derive the mapping between
|
| 108 |
+
the two layers from frame-interval overlap instead. This dataset carries exactly one
|
| 109 |
+
`episode_subtasks` row per episode (= the whole-episode task).
|
example_data/umi_data_sample/annotation/action_steps_000001.jsonl
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
-
{"subtask_index":
|
| 2 |
-
{"subtask_index":
|
| 3 |
-
{"subtask_index":
|
| 4 |
-
{"subtask_index":
|
|
|
|
| 1 |
+
{"subtask_index": 0, "start_frame_index": 0, "end_frame_index": 45, "instruction": "grasp the black shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 2 |
+
{"subtask_index": 0, "start_frame_index": 45, "end_frame_index": 135, "instruction": "lay the black shirt flat", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 3 |
+
{"subtask_index": 0, "start_frame_index": 135, "end_frame_index": 330, "instruction": "smooth out wrinkles on the black shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 4 |
+
{"subtask_index": 0, "start_frame_index": 390, "end_frame_index": 1050, "instruction": "press the black shirt to remove wrinkles", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
example_data/umi_data_sample/annotation/action_steps_000002.jsonl
CHANGED
|
@@ -1,18 +1,18 @@
|
|
| 1 |
-
{"subtask_index":
|
| 2 |
-
{"subtask_index":
|
| 3 |
-
{"subtask_index":
|
| 4 |
-
{"subtask_index":
|
| 5 |
-
{"subtask_index":
|
| 6 |
-
{"subtask_index":
|
| 7 |
-
{"subtask_index":
|
| 8 |
-
{"subtask_index":
|
| 9 |
-
{"subtask_index":
|
| 10 |
-
{"subtask_index":
|
| 11 |
-
{"subtask_index":
|
| 12 |
-
{"subtask_index":
|
| 13 |
-
{"subtask_index":
|
| 14 |
-
{"subtask_index":
|
| 15 |
-
{"subtask_index":
|
| 16 |
-
{"subtask_index":
|
| 17 |
-
{"subtask_index":
|
| 18 |
-
{"subtask_index":
|
|
|
|
| 1 |
+
{"subtask_index": 0, "start_frame_index": 0, "end_frame_index": 30, "instruction": "reach for the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 2 |
+
{"subtask_index": 0, "start_frame_index": 30, "end_frame_index": 60, "instruction": "grasp the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 3 |
+
{"subtask_index": 0, "start_frame_index": 60, "end_frame_index": 105, "instruction": "lift and move the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 4 |
+
{"subtask_index": 0, "start_frame_index": 105, "end_frame_index": 210, "instruction": "lay the yellow shirt flat", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 5 |
+
{"subtask_index": 0, "start_frame_index": 210, "end_frame_index": 270, "instruction": "smooth out wrinkles on the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 6 |
+
{"subtask_index": 0, "start_frame_index": 270, "end_frame_index": 330, "instruction": "reach for the bottom edge of the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 7 |
+
{"subtask_index": 0, "start_frame_index": 330, "end_frame_index": 375, "instruction": "fold the bottom edge of the yellow shirt up", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 8 |
+
{"subtask_index": 0, "start_frame_index": 375, "end_frame_index": 435, "instruction": "smooth out wrinkles on the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 9 |
+
{"subtask_index": 0, "start_frame_index": 435, "end_frame_index": 480, "instruction": "reach for the left side of the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 10 |
+
{"subtask_index": 0, "start_frame_index": 480, "end_frame_index": 540, "instruction": "fold the left side of the yellow shirt over", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 11 |
+
{"subtask_index": 0, "start_frame_index": 540, "end_frame_index": 585, "instruction": "smooth out wrinkles on the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 12 |
+
{"subtask_index": 0, "start_frame_index": 585, "end_frame_index": 630, "instruction": "reach for the right side of the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 13 |
+
{"subtask_index": 0, "start_frame_index": 630, "end_frame_index": 675, "instruction": "fold the right side of the yellow shirt over", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 14 |
+
{"subtask_index": 0, "start_frame_index": 675, "end_frame_index": 720, "instruction": "smooth out wrinkles on the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 15 |
+
{"subtask_index": 0, "start_frame_index": 720, "end_frame_index": 765, "instruction": "reach for the bottom edge of the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 16 |
+
{"subtask_index": 0, "start_frame_index": 765, "end_frame_index": 810, "instruction": "fold the yellow shirt in half", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 17 |
+
{"subtask_index": 0, "start_frame_index": 810, "end_frame_index": 855, "instruction": "place/stack the yellow shirt on the pile", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 18 |
+
{"subtask_index": 0, "start_frame_index": 855, "end_frame_index": 870, "instruction": "release the yellow shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
example_data/umi_data_sample/annotation/action_steps_000003.jsonl
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
-
{"subtask_index":
|
| 2 |
-
{"subtask_index":
|
| 3 |
-
{"subtask_index":
|
| 4 |
-
{"subtask_index":
|
| 5 |
-
{"subtask_index":
|
| 6 |
-
{"subtask_index":
|
| 7 |
-
{"subtask_index":
|
| 8 |
-
{"subtask_index":
|
|
|
|
| 1 |
+
{"subtask_index": 0, "start_frame_index": 0, "end_frame_index": 180, "instruction": "lay the black shirt flat", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 2 |
+
{"subtask_index": 0, "start_frame_index": 180, "end_frame_index": 270, "instruction": "smooth out wrinkles on the black shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 3 |
+
{"subtask_index": 0, "start_frame_index": 300, "end_frame_index": 360, "instruction": "fold one side of the black shirt over", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 4 |
+
{"subtask_index": 0, "start_frame_index": 390, "end_frame_index": 495, "instruction": "fold the other side of the black shirt over", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 5 |
+
{"subtask_index": 0, "start_frame_index": 585, "end_frame_index": 660, "instruction": "fold the bottom of the black shirt up", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 6 |
+
{"subtask_index": 0, "start_frame_index": 675, "end_frame_index": 795, "instruction": "smooth out wrinkles on the black shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 7 |
+
{"subtask_index": 0, "start_frame_index": 810, "end_frame_index": 915, "instruction": "grasp the black shirt", "instruction_augmentation": [], "actor": "both", "actor_confidence": "low", "success": true}
|
| 8 |
+
{"subtask_index": 0, "start_frame_index": 930, "end_frame_index": 1050, "instruction": "fold the black shirt in half", "instruction_augmentation": [], "actor": "both", "actor_confidence": "low", "success": true}
|
example_data/umi_data_sample/annotation/action_steps_000004.jsonl
CHANGED
|
@@ -1,27 +1,27 @@
|
|
| 1 |
-
{"subtask_index":
|
| 2 |
-
{"subtask_index":
|
| 3 |
-
{"subtask_index":
|
| 4 |
-
{"subtask_index":
|
| 5 |
-
{"subtask_index":
|
| 6 |
-
{"subtask_index":
|
| 7 |
-
{"subtask_index":
|
| 8 |
-
{"subtask_index":
|
| 9 |
-
{"subtask_index":
|
| 10 |
-
{"subtask_index":
|
| 11 |
-
{"subtask_index":
|
| 12 |
-
{"subtask_index":
|
| 13 |
-
{"subtask_index":
|
| 14 |
-
{"subtask_index":
|
| 15 |
-
{"subtask_index":
|
| 16 |
-
{"subtask_index":
|
| 17 |
-
{"subtask_index":
|
| 18 |
-
{"subtask_index":
|
| 19 |
-
{"subtask_index":
|
| 20 |
-
{"subtask_index":
|
| 21 |
-
{"subtask_index":
|
| 22 |
-
{"subtask_index":
|
| 23 |
-
{"subtask_index":
|
| 24 |
-
{"subtask_index":
|
| 25 |
-
{"subtask_index":
|
| 26 |
-
{"subtask_index":
|
| 27 |
-
{"subtask_index":
|
|
|
|
| 1 |
+
{"subtask_index": 0, "start_frame_index": 75, "end_frame_index": 90, "instruction": "reach for the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 2 |
+
{"subtask_index": 0, "start_frame_index": 90, "end_frame_index": 105, "instruction": "grasp the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 3 |
+
{"subtask_index": 0, "start_frame_index": 105, "end_frame_index": 150, "instruction": "lift the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 4 |
+
{"subtask_index": 0, "start_frame_index": 150, "end_frame_index": 195, "instruction": "lay the brown shirt flat", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 5 |
+
{"subtask_index": 0, "start_frame_index": 195, "end_frame_index": 645, "instruction": "smooth out wrinkles on the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 6 |
+
{"subtask_index": 0, "start_frame_index": 660, "end_frame_index": 690, "instruction": "reach for one side of the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 7 |
+
{"subtask_index": 0, "start_frame_index": 690, "end_frame_index": 705, "instruction": "grasp it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 8 |
+
{"subtask_index": 0, "start_frame_index": 705, "end_frame_index": 720, "instruction": "lift it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 9 |
+
{"subtask_index": 0, "start_frame_index": 720, "end_frame_index": 750, "instruction": "fold one side over", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 10 |
+
{"subtask_index": 0, "start_frame_index": 750, "end_frame_index": 840, "instruction": "smooth out wrinkles on the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 11 |
+
{"subtask_index": 0, "start_frame_index": 840, "end_frame_index": 870, "instruction": "reach for the other side of the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 12 |
+
{"subtask_index": 0, "start_frame_index": 870, "end_frame_index": 885, "instruction": "grasp it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 13 |
+
{"subtask_index": 0, "start_frame_index": 885, "end_frame_index": 900, "instruction": "lift it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 14 |
+
{"subtask_index": 0, "start_frame_index": 900, "end_frame_index": 930, "instruction": "fold the other side over", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 15 |
+
{"subtask_index": 0, "start_frame_index": 930, "end_frame_index": 1125, "instruction": "smooth out wrinkles on the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 16 |
+
{"subtask_index": 0, "start_frame_index": 1125, "end_frame_index": 1155, "instruction": "reach for the bottom of the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 17 |
+
{"subtask_index": 0, "start_frame_index": 1155, "end_frame_index": 1170, "instruction": "grasp it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 18 |
+
{"subtask_index": 0, "start_frame_index": 1170, "end_frame_index": 1185, "instruction": "lift it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 19 |
+
{"subtask_index": 0, "start_frame_index": 1185, "end_frame_index": 1215, "instruction": "fold the brown shirt in half", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 20 |
+
{"subtask_index": 0, "start_frame_index": 1215, "end_frame_index": 1350, "instruction": "smooth out wrinkles on the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 21 |
+
{"subtask_index": 0, "start_frame_index": 1350, "end_frame_index": 1380, "instruction": "reach for the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 22 |
+
{"subtask_index": 0, "start_frame_index": 1380, "end_frame_index": 1395, "instruction": "grasp it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 23 |
+
{"subtask_index": 0, "start_frame_index": 1395, "end_frame_index": 1410, "instruction": "lift it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 24 |
+
{"subtask_index": 0, "start_frame_index": 1410, "end_frame_index": 1455, "instruction": "fold the brown shirt in half again", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 25 |
+
{"subtask_index": 0, "start_frame_index": 1455, "end_frame_index": 1515, "instruction": "smooth out wrinkles on the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 26 |
+
{"subtask_index": 0, "start_frame_index": 1515, "end_frame_index": 1545, "instruction": "reach for the brown shirt", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
| 27 |
+
{"subtask_index": 0, "start_frame_index": 1545, "end_frame_index": 1560, "instruction": "grasp it", "instruction_augmentation": [], "actor": "right", "actor_confidence": "low", "success": true}
|
example_data/umi_data_sample/data/chunk-000/episode_000000.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1f5522d5f5f66a01f1759a0e82eb9b74b7dc3a9c1e61416408efb37c0559ac1
|
| 3 |
+
size 267374
|
example_data/umi_data_sample/data/chunk-000/episode_000001.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:679dad3225563f18a9b0b914e79aa39ec433ea57e79938b6abc394a68671c9e7
|
| 3 |
+
size 200399
|
example_data/umi_data_sample/data/chunk-000/episode_000002.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e3868f76c3423611c7478a7698dfd3e1425bc019bfaa2804c1cddeaf50e6e1f0
|
| 3 |
+
size 162284
|
example_data/umi_data_sample/data/chunk-000/episode_000003.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7787d2d732f42aaa68b2cc4ad442169d82c83b60f5c1d720ecc9cd6ae8c05321
|
| 3 |
+
size 198963
|
example_data/umi_data_sample/data/chunk-000/episode_000004.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4751e72e815d043770798412a422f6e4f27ba37d252ae5c6d9e643ba9d9c5a64
|
| 3 |
+
size 293824
|
example_data/umi_data_sample/meta/calibration.json
CHANGED
|
@@ -135,7 +135,7 @@
|
|
| 135 |
}
|
| 136 |
}
|
| 137 |
},
|
| 138 |
-
"
|
| 139 |
"R_cam_imu": [
|
| 140 |
[
|
| 141 |
0.0029496021588553776,
|
|
@@ -164,7 +164,7 @@
|
|
| 164 |
"camera_intrinsics": "Factory per-camera telemetry",
|
| 165 |
"kalibr_refined": false,
|
| 166 |
"pose_frame": "cubemap front virtual pinhole, 110 deg FoV / tilt 70 / 640 px",
|
| 167 |
-
"pose_frame_note": "state/action poses (T_w_cam) live in this virtual pinhole frame, NOT the fisheye frame of the delivered video. When
|
| 168 |
}
|
| 169 |
},
|
| 170 |
"right": {
|
|
@@ -301,7 +301,7 @@
|
|
| 301 |
}
|
| 302 |
}
|
| 303 |
},
|
| 304 |
-
"
|
| 305 |
"R_cam_imu": [
|
| 306 |
[
|
| 307 |
-0.0018151414253366012,
|
|
@@ -330,7 +330,7 @@
|
|
| 330 |
"camera_intrinsics": "Factory per-camera telemetry",
|
| 331 |
"kalibr_refined": false,
|
| 332 |
"pose_frame": "cubemap front virtual pinhole, 110 deg FoV / tilt 70 / 640 px",
|
| 333 |
-
"pose_frame_note": "state/action poses (T_w_cam) live in this virtual pinhole frame, NOT the fisheye frame of the delivered video. When
|
| 334 |
}
|
| 335 |
}
|
| 336 |
}
|
|
|
|
| 135 |
}
|
| 136 |
}
|
| 137 |
},
|
| 138 |
+
"cam_imu_extrinsics": {
|
| 139 |
"R_cam_imu": [
|
| 140 |
[
|
| 141 |
0.0029496021588553776,
|
|
|
|
| 164 |
"camera_intrinsics": "Factory per-camera telemetry",
|
| 165 |
"kalibr_refined": false,
|
| 166 |
"pose_frame": "cubemap front virtual pinhole, 110 deg FoV / tilt 70 / 640 px",
|
| 167 |
+
"pose_frame_note": "state/action poses (T_w_cam) live in this virtual pinhole frame, NOT the fisheye frame of the delivered video. When a body has no Kalibr-refined camera-IMU calibration the lookup table is borrowed from another body of the same model (see provenance.kalibr_refined)."
|
| 168 |
}
|
| 169 |
},
|
| 170 |
"right": {
|
|
|
|
| 301 |
}
|
| 302 |
}
|
| 303 |
},
|
| 304 |
+
"cam_imu_extrinsics": {
|
| 305 |
"R_cam_imu": [
|
| 306 |
[
|
| 307 |
-0.0018151414253366012,
|
|
|
|
| 330 |
"camera_intrinsics": "Factory per-camera telemetry",
|
| 331 |
"kalibr_refined": false,
|
| 332 |
"pose_frame": "cubemap front virtual pinhole, 110 deg FoV / tilt 70 / 640 px",
|
| 333 |
+
"pose_frame_note": "state/action poses (T_w_cam) live in this virtual pinhole frame, NOT the fisheye frame of the delivered video. When a body has no Kalibr-refined camera-IMU calibration the lookup table is borrowed from another body of the same model (see provenance.kalibr_refined)."
|
| 334 |
}
|
| 335 |
}
|
| 336 |
}
|