File size: 2,833 Bytes
2e04839 e39653f 2e04839 e39653f 2e04839 e39653f 2e04839 e39653f 2e04839 e39653f 2e04839 e39653f 2e04839 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 | ---
license: other
task_categories:
- robotics
- video-classification
tags:
- imitation-learning
- mujoco
- robot-arm
- robotics
- so101
- grasping
pretty_name: SO-101 Color-Split Grasp Trajectory Dataset
---
# SO-101 Color-Split Grasp Trajectory Dataset
This dataset contains 540 SO-101 robot-arm block grasp trajectories generated in MuJoCo and split by target block color.
- blue: 180 trajectories
- yellow: 180 trajectories
- red: 180 trajectories
Each color subset contains:
- `trajectory_summary.csv`: per-trajectory metadata and success fields
- `joint_trajectories.csv`: per-step joint states and action commands for learning
- `batch_summary.json`: subset summary
- `index.html`: visual index for the color subset
- `*_camera1_sequence.mp4`, `*_camera2_sequence.mp4`, `*_camera3_sequence.mp4`: concatenated visualization videos
All rows in the prepared package have `grasp_lifted=True`.
## Learning Columns
Use `joint_trajectories.csv` for robot learning. Each row is one simulation step in one trajectory.
- State joints: `actual_shoulder_pan_deg`, `actual_shoulder_lift_deg`, `actual_elbow_flex_deg`, `actual_wrist_flex_deg`, `actual_wrist_roll_deg`, `actual_gripper_deg`
- Action/command joints: `target_shoulder_pan_deg`, `target_shoulder_lift_deg`, `target_elbow_flex_deg`, `target_wrist_flex_deg`, `target_wrist_roll_deg`, `target_gripper_deg`
- Command error: `delta_*_deg = target_*_deg - actual_*_deg`
- Extra context: `phase`, `target_color`, `style`, `target_x`, `target_y`, target block position, jaw center, fingertip positions, reward, distance, lift height, contact flags
The dataset includes 363,288 per-step joint rows in total: 121,096 rows for each target color.
## Layout
```text
data/
blue/
joint_trajectories.csv
yellow/
joint_trajectories.csv
red/
joint_trajectories.csv
dataset_summary.json
joint_trajectory_summary.json
manifest.csv
```
## Generation Summary
```json
{
"source_batch_name": "20260603_031736_three_target_colors_line120_y220_target_only_robot_lr30mm_angles5_permutations_540",
"total_trajectories": 540,
"joint_trajectory_rows": 363288,
"joint_columns": {
"state": "actual_*_deg",
"action_command": "target_*_deg",
"error": "delta_*_deg",
"joints": [
"shoulder_pan",
"shoulder_lift",
"elbow_flex",
"wrist_flex",
"wrist_roll",
"gripper"
]
},
"colors": {
"blue": {
"trajectories": 180,
"grasp_lifted": 180,
"joint_trajectory_rows": 121096,
"path": "data/blue"
},
"yellow": {
"trajectories": 180,
"grasp_lifted": 180,
"joint_trajectory_rows": 121096,
"path": "data/yellow"
},
"red": {
"trajectories": 180,
"grasp_lifted": 180,
"joint_trajectory_rows": 121096,
"path": "data/red"
}
}
}
```
|