Sraghvi commited on
Commit
8d9c8ee
·
verified ·
1 Parent(s): 0f2a6f1

Upload 4 files

Browse files
Files changed (4) hide show
  1. README.md +181 -0
  2. data-00000-of-00001.arrow +3 -0
  3. dataset_info.json +44 -0
  4. state.json +13 -0
README.md ADDED
@@ -0,0 +1,181 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ task_categories:
4
+ - robotics
5
+ tags:
6
+ - LeRobot
7
+ - multi-robot
8
+ - manipulation
9
+ - bone-picking
10
+ configs:
11
+ - config_name: default
12
+ data_files: data/*/*.parquet
13
+ ---
14
+
15
+ # Bone to Pick - Combined Dataset (10_08 + 10_09)
16
+
17
+ This dataset is a merged combination of two Bone to Pick datasets from Hugging Face:
18
+ - [oordonez/Bone_to_Pick_10_08](https://huggingface.co/datasets/oordonez/Bone_to_Pick_10_08)
19
+ - [oordonez/Bone_to_Pick_10_09](https://huggingface.co/datasets/oordonez/Bone_to_Pick_10_09)
20
+
21
+ ## Dataset Description
22
+
23
+ This dataset was created using [LeRobot](https://github.com/huggingface/lerobot) and contains bi-manual robot manipulation data for a bone picking task. The dataset combines 20 episodes (10 from each original dataset) with updated episode numbering to avoid conflicts.
24
+
25
+ ### Key Features:
26
+ - **Total Episodes**: 20 (episodes 0-19)
27
+ - **Total Frames**: 7,788 (after dead time removal)
28
+ - **Total Videos**: 60 (3 camera views × 20 episodes)
29
+ - **Robot Type**: Dual-arm manipulation (white + blue follower arms)
30
+ - **Task**: Grasp the blue bone and place it inside the box
31
+
32
+ ### Camera Views:
33
+ 1. `observation.images.my_blue_follower_arm_front` (640×480, H.264)
34
+ 2. `observation.images.my_white_follower_arm_front` (640×480, H.264)
35
+ 3. `observation.images.my_white_follower_arm_global_top` (1280×720, H.264)
36
+
37
+ ## Dataset Structure
38
+
39
+ ### Data Format:
40
+ - **Action Space**: 12-dimensional float32 arrays
41
+ - 6 joints for white follower arm
42
+ - 6 joints for blue follower arm
43
+ - **Observation Space**: 12-dimensional float32 arrays (same as action)
44
+ - **Timestamps**: float32 timestamps
45
+ - **Frame Indices**: int64 frame numbers within each episode
46
+ - **Episode Indices**: int64 episode numbers (0-19)
47
+
48
+ ### Joint Names:
49
+ **White Follower Arm:**
50
+ - `my_white_follower_arm_shoulder_pan.pos`
51
+ - `my_white_follower_arm_shoulder_lift.pos`
52
+ - `my_white_follower_arm_elbow_flex.pos`
53
+ - `my_white_follower_arm_wrist_flex.pos`
54
+ - `my_white_follower_arm_wrist_roll.pos`
55
+ - `my_white_follower_arm_gripper.pos`
56
+
57
+ **Blue Follower Arm:**
58
+ - `my_blue_follower_arm_shoulder_pan.pos`
59
+ - `my_blue_follower_arm_shoulder_lift.pos`
60
+ - `my_blue_follower_arm_elbow_flex.pos`
61
+ - `my_blue_follower_arm_wrist_flex.pos`
62
+ - `my_blue_follower_arm_wrist_roll.pos`
63
+ - `my_blue_follower_arm_gripper.pos`
64
+
65
+ ## Processing Details
66
+
67
+ ### Dead Time Removal:
68
+ - Applied velocity threshold of 0.0001 for joint movement detection
69
+ - Removed 2+ seconds of stationary data at episode ends
70
+ - Total frames removed: 1,496 frames across all episodes
71
+
72
+ ### Video Processing:
73
+ - Converted all videos from AV1 to H.264 codec for browser compatibility
74
+ - Maintained original frame rates (30 FPS)
75
+ - Preserved video quality and aspect ratios
76
+
77
+ ### Episode Renumbering:
78
+ - Original dataset 10_08: episodes 0-9 → merged episodes 0-9
79
+ - Original dataset 10_09: episodes 0-9 → merged episodes 10-19
80
+ - All metadata updated to reflect new episode numbering
81
+
82
+ ## Dataset Viewer Information
83
+
84
+ ### What You'll See in the Hugging Face Dataset Viewer:
85
+
86
+ #### **📊 Data Columns:**
87
+ - **`action`**: 12-element array containing joint positions for both robot arms
88
+ - **`observation.state`**: 12-element array containing current joint states
89
+ - **`timestamp`**: float32 timestamp for each frame
90
+ - **`frame_index`**: int64 frame number within each episode
91
+ - **`episode_index`**: int64 episode number (0-19)
92
+ - **`index`**: int64 global sample index
93
+ - **`task_index`**: int64 task identifier
94
+
95
+ #### **🤖 Joint Mapping (12 elements in action/observation arrays):**
96
+ **Indices 0-5: White Follower Arm**
97
+ - Index 0: `my_white_follower_arm_shoulder_pan.pos`
98
+ - Index 1: `my_white_follower_arm_shoulder_lift.pos`
99
+ - Index 2: `my_white_follower_arm_elbow_flex.pos`
100
+ - Index 3: `my_white_follower_arm_wrist_flex.pos`
101
+ - Index 4: `my_white_follower_arm_wrist_roll.pos`
102
+ - Index 5: `my_white_follower_arm_gripper.pos`
103
+
104
+ **Indices 6-11: Blue Follower Arm**
105
+ - Index 6: `my_blue_follower_arm_shoulder_pan.pos`
106
+ - Index 7: `my_blue_follower_arm_shoulder_lift.pos`
107
+ - Index 8: `my_blue_follower_arm_elbow_flex.pos`
108
+ - Index 9: `my_blue_follower_arm_wrist_flex.pos`
109
+ - Index 10: `my_blue_follower_arm_wrist_roll.pos`
110
+ - Index 11: `my_blue_follower_arm_gripper.pos`
111
+
112
+ #### **🎥 Video Files:**
113
+ - **60 total videos** (3 camera views × 20 episodes)
114
+ - All videos are in H.264 codec for browser compatibility
115
+ - Videos are stored in the `videos/` directory
116
+
117
+ ## Usage
118
+
119
+ ### Loading the Dataset:
120
+ ```python
121
+ from datasets import load_from_disk
122
+
123
+ # Load the dataset
124
+ dataset = load_from_disk('merged_bone_to_pick_combined_hf_fixed')
125
+
126
+ # Access a sample
127
+ sample = dataset[0]
128
+ print(f"Action: {sample['action']}")
129
+ print(f"Observation: {sample['observation.state']}")
130
+ print(f"Episode: {sample['episode_index']}")
131
+
132
+ # Access joint values by index
133
+ action = sample['action']
134
+ white_arm_joints = action[:6] # First 6 elements
135
+ blue_arm_joints = action[6:] # Last 6 elements
136
+ ```
137
+
138
+ ### Accessing Videos:
139
+ ```python
140
+ # Videos are stored in the videos/ directory
141
+ # Each episode has 3 camera views:
142
+ # - observation.images.my_blue_follower_arm_front/episode_XXXXXX.mp4
143
+ # - observation.images.my_white_follower_arm_front/episode_XXXXXX.mp4
144
+ # - observation.images.my_white_follower_arm_global_top/episode_XXXXXX.mp4
145
+ ```
146
+
147
+ ## Dataset Statistics
148
+
149
+ | Metric | Value |
150
+ |--------|-------|
151
+ | Total Episodes | 20 |
152
+ | Total Frames | 7,788 |
153
+ | Total Videos | 60 |
154
+ | Average Frames per Episode | 389.4 |
155
+ | Video Codec | H.264 |
156
+ | Frame Rate | 30 FPS |
157
+ | Data Format | LeRobot v2.1 |
158
+
159
+ ## Citation
160
+
161
+ **BibTeX:**
162
+ ```bibtex
163
+ @dataset{merged_bone_to_pick_combined,
164
+ title={Bone to Pick - Combined Dataset (10_08 + 10_09)},
165
+ author={oordonez},
166
+ year={2024},
167
+ publisher={Hugging Face},
168
+ url={https://huggingface.co/datasets/oordonez/Bone_to_Pick_10_08},
169
+ url={https://huggingface.co/datasets/oordonez/Bone_to_Pick_10_09}
170
+ }
171
+ ```
172
+
173
+ ## License
174
+
175
+ This dataset is released under the Apache 2.0 License.
176
+
177
+ ## Acknowledgments
178
+
179
+ - Original datasets: [oordonez/Bone_to_Pick_10_08](https://huggingface.co/datasets/oordonez/Bone_to_Pick_10_08) and [oordonez/Bone_to_Pick_10_09](https://huggingface.co/datasets/oordonez/Bone_to_Pick_10_09)
180
+ - Built with [LeRobot](https://github.com/huggingface/lerobot)
181
+ - Merged and processed for improved usability and browser compatibility
data-00000-of-00001.arrow ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e3cca359c1c54bbb0958acdaefbfd549e862dde46ae18867526d8191855e71a
3
+ size 1033104
dataset_info.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "citation": "",
3
+ "description": "",
4
+ "features": {
5
+ "action": {
6
+ "feature": {
7
+ "dtype": "float32",
8
+ "_type": "Value"
9
+ },
10
+ "length": 12,
11
+ "_type": "List"
12
+ },
13
+ "observation.state": {
14
+ "feature": {
15
+ "dtype": "float32",
16
+ "_type": "Value"
17
+ },
18
+ "length": 12,
19
+ "_type": "List"
20
+ },
21
+ "timestamp": {
22
+ "dtype": "float32",
23
+ "_type": "Value"
24
+ },
25
+ "frame_index": {
26
+ "dtype": "int64",
27
+ "_type": "Value"
28
+ },
29
+ "episode_index": {
30
+ "dtype": "int64",
31
+ "_type": "Value"
32
+ },
33
+ "index": {
34
+ "dtype": "int64",
35
+ "_type": "Value"
36
+ },
37
+ "task_index": {
38
+ "dtype": "int64",
39
+ "_type": "Value"
40
+ }
41
+ },
42
+ "homepage": "",
43
+ "license": ""
44
+ }
state.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_data_files": [
3
+ {
4
+ "filename": "data-00000-of-00001.arrow"
5
+ }
6
+ ],
7
+ "_fingerprint": "1556691ee348eb7e",
8
+ "_format_columns": null,
9
+ "_format_kwargs": {},
10
+ "_format_type": null,
11
+ "_output_all_columns": false,
12
+ "_split": null
13
+ }