merged_rl_lv2emp

Merged LeRobot dataset (SO-101 cube pick-and-place) — 250 episodes / 83,984 frames. Concatenation of the source datasets below.

Source datasets

Source	Episodes
geonmin-kim/SO101-lv2-single-cube-to-box-random-layout-v1	200
geonmin-kim/lv2_neartop_emptyupbox	50
Total	250

Tasks

• "Pick up the {color} cube ..." — 5 colors (from random-layout-v1)
• "Pick up any {color} cube ..." — 5 colors (from lv2_neartop_emptyupbox)

Note: the merged task table keeps each source's original instruction phrasing verbatim, so semantically-equivalent tasks may appear under multiple strings.

Format

• Robot: SO-101 (so_follower), 6-DOF joint-position control (shoulder_pan, shoulder_lift, elbow_flex, wrist_flex, wrist_roll, gripper).
• Cameras: observation.images.top, observation.images.wrist (480×640×3, mp4).
• Rate: 30 fps. Format: LeRobot dataset v3.0.
• Built with lerobot.datasets.dataset_tools.merge_datasets (episodes concatenated; task table unioned).

Usage

from lerobot.datasets.lerobot_dataset import LeRobotDataset
ds = LeRobotDataset("geonmin-kim/merged_rl_lv2emp")