Datasetpublic

upna/bimanual-pass-double-sided-tape-roll

bimanual_handover_4cam_20260717_150424 Bimanual SO-101 teleoperation dataset for a double-sided tape handover task, recorded with 4 cameras in LeRobot format (v3.0). This is a 4-camera collection (distinct from the older 3-camera upna/bimanual_handover). Task Grasp the double-sided tape roll, hand it over to the other arm, and place it in the yellow rectangular plastic tray. Dataset summary Field Value Robot bi_so101_follower (dual… See the full description on the dataset page: https://huggingface.co/datasets/upna/bimanual-pass-double-sided-tape-roll.

sourceHugging Faceapache-2.0updated 2mo agoView on Hugging Face

1likes63downloads

Dataset Card

bimanualhandover4cam20260717150424

Bimanual SO-101 teleoperation dataset for a double-sided tape handover task, recorded with 4 cameras in LeRobot format (v3.0).

This is a 4-camera collection (distinct from the older 3-camera upna/bimanual_handover).

Task

Grasp the double-sided tape roll, hand it over to the other arm, and place it in the yellow rectangular plastic tray.

Dataset summary

Field	Value
Robot	`bi_so101_follower` (dual SO-101 arms)
Collection	Leader–follower teleoperation
Episodes	89
Frames	47,074
FPS	30
Episode length	~413–680 frames (~13.8–22.7 s), mean ~529 frames (~17.6 s)
Tasks	1
Split	`train`: episodes `0:89`
Format	LeRobot codebase `v3.0`
Approx. size	~480–500 MB

Cameras

Key	Resolution	Role
`observation.images.top_left`	800 × 480	Overhead / left-top view
`observation.images.top_right`	640 × 480	Overhead / right-top view
`observation.images.left`	640 × 480	Left wrist camera
`observation.images.right`	640 × 480	Right wrist camera

All videos: H.264, 30 FPS, RGB, no audio.

Action / state

12-DoF joint positions (left arm 6 + right arm 6):

—*_shoulder_pan.pos
—*_shoulder_lift.pos
—*_elbow_flex.pos
—*_wrist_flex.pos
—*_wrist_roll.pos
—*_gripper.pos

Features:

—action — commanded / teleop joint targets (float32, shape [12])
—observation.state — measured joint positions (float32, shape [12])

Features

—observation.images.top_left / top_right / left / right (video)
—observation.state
—action
—timestamp, frame_index, episode_index, index, task_index

Load with LeRobot

python

from lerobot.datasets.lerobot_dataset import LeRobotDataset

dataset = LeRobotDataset(
    repo_id="upna/bimanual_handover_4cam_20260717_150424",  # change if your Hub id differs
    # root="/home/rxn/datasets/bimanual_handover_4cam_20260717_150424",  # local path
)

print(dataset.meta)
print(len(dataset), "frames,", dataset.num_episodes, "episodes")

Visualize locally:

bash

lerobot-dataset-viz \
  --repo-id upna/bimanual_handover_4cam_20260717_150424 \
  --root /home/rxn/datasets/bimanual_handover_4cam_20260717_150424 \
  --episode-index 0

Collection notes

—Recorded with LeRobot bimanual SO-101 leader–follower teleop (get-data-bimanual.sh).
—Local root: /home/rxn/datasets/bimanual_handover_4cam_20260717_150424
—Timestamp suffix 20260717_150424 identifies this collection run.
—Compared to 3-cam bimanual_handover, this run adds top_left (800×480) alongside top_right / wrist cams.

Citation

If you use this dataset, please cite LeRobot and this repository.