Add dataset card
Browse files
README.md
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
tags: [LeRobot, robotics, manipulation, imitation-learning]
|
| 4 |
+
task_categories: [robotics]
|
| 5 |
+
size_categories: [1K<n<10K]
|
| 6 |
+
---
|
| 7 |
+
|
| 8 |
+
# datagen-jar-v1-joint-5cam
|
| 9 |
+
|
| 10 |
+
Auto-generated SFT dataset for the **jar_transport** family (close an articulated hinged jar's lid,
|
| 11 |
+
then side-grasp the closed jar and carry it to a goal region) — cuRobo-planned, physics- &
|
| 12 |
+
LTL-safety-checked demos, converted to **LeRobot v2.1**.
|
| 13 |
+
|
| 14 |
+
- **Creator:** yypeng666 (IDEAS-Lab-Northwestern)
|
| 15 |
+
- **Source bench:** [IDEAS-Lab-Northwestern/ManiGuard-Bench](https://huggingface.co/datasets/IDEAS-Lab-Northwestern/ManiGuard-Bench) — all 26 `jar_transport` base tasks.
|
| 16 |
+
- **Per task:** 40 success + LTL-safe trajectories → **1040 episodes**. Visibility: private.
|
| 17 |
+
|
| 18 |
+
## Contents
|
| 19 |
+
| | |
|
| 20 |
+
|---|---|
|
| 21 |
+
| Episodes | 1040 (26 base tasks × 40) |
|
| 22 |
+
| Frames | 946,870 |
|
| 23 |
+
| Unique language tasks | 21 |
|
| 24 |
+
| Robot / control | FrankaPanda / absolute joint |
|
| 25 |
+
| FPS / resolution / format | 30 / 256×256 / LeRobot v2.1 |
|
| 26 |
+
|
| 27 |
+
## Task
|
| 28 |
+
The robot closes an articulated hinged jar's lid — one straight "lid-ride" push of the open gripper
|
| 29 |
+
under the flopped lid, lifting it past its tipping point so gravity seats it shut — then side-grasps
|
| 30 |
+
the closed jar and transports it into the green goal sphere. Per-episode prompt (`meta/tasks.jsonl`)
|
| 31 |
+
from the base task diagnostics. A demo must reach the goal AND stay LTL-safe throughout (jar upright,
|
| 32 |
+
not dropped, on its support until closed); failed / unsafe attempts are dropped.
|
| 33 |
+
|
| 34 |
+
## Schema
|
| 35 |
+
`state`(8)=`[arm_q(7),gripper(1)]`; `actions`(8)=`[arm_q[t+1](7),gripper_cmd(1)]` next-achieved
|
| 36 |
+
(**default SFT target**); `actions_commanded`(8)=`[curobo_target_q(7),gripper_cmd(1)]` (cuRobo command,
|
| 37 |
+
extra). Absolute joint (delta at train time). Cameras (all 5 kept, pick subset at train time):
|
| 38 |
+
`image_opposite,image_left,image_right,image_left_shoulder` (third-person) + `wrist_image`.
|
| 39 |
+
|
| 40 |
+
## Notes
|
| 41 |
+
- Camera subset + arm-joint delta are TRAIN-time choices; all 5 streams + absolute joints ship here.
|
| 42 |
+
- MimicGen sim-state replay dump is NOT included (kept in the raw archive; not needed for SFT).
|