XYZPIT/260908_pick_up_the_specified_bread_bag

260908 Pick Up the Specified Bread Bag

Dataset Description

This dataset contains teleoperated demonstrations collected with the Galaxea R1 Lite robot. Three bread bags are placed at the left, middle, and right positions. In each episode, the robot picks up the specified bread bag using the specified hand.

The demonstrations contain only the picking motion. They do not include a subsequent placing task.

The dataset is stored in the LeRobot dataset format and contains six task instructions based on the target position and the hand used.

Dataset Summary

Item	Value
Robot	Galaxea R1 Lite
Collection method	Teleoperation
Number of episodes	140
Number of frames	44,315
Number of videos	560
Number of task instructions	6
Frames per second	31 FPS

Task Distribution

Task instruction	Episodes
Pick up the left bread bag with the left hand.	20
Pick up the left bread bag with the right hand.	20
Pick up the middle bread bag with the left hand.	30
Pick up the middle bread bag with the right hand.	30
Pick up the right bread bag with the left hand.	20
Pick up the right bread bag with the right hand.	20
Total	140

Observations

Each episode contains four synchronized RGB camera streams:

Feature	View	Resolution
observation.images.head_rgb	Left head camera	1280 × 720
observation.images.head_right_rgb	Right head camera	1280 × 720
observation.images.left_wrist_rgb	Left wrist camera	640 × 360
observation.images.right_wrist_rgb	Right wrist camera	640 × 360

The videos are encoded using AV1 at 31 FPS. The dataset also includes robot proprioceptive observations such as arm joint positions and velocities, gripper states, end-effector poses, chassis and torso states, and IMU measurements.

Actions

The action features include:

• Left and right arm joint targets
• Left and right gripper targets
• Chassis velocity commands
• Torso velocity commands

Intended Use

This dataset is intended for research on robot imitation learning and language-conditioned manipulation, particularly for learning to select an object by spatial position and execute the pickup using a specified hand.

Limitations

• The demonstrations were collected in a single workspace with three bread bags arranged at predefined left, middle, and right positions.
• The middle-position tasks contain more episodes than the left- and right-position tasks.
• The dataset covers picking only and does not include object placement.
• Performance learned from this dataset may not directly generalize to substantially different objects, layouts, lighting conditions, or robot embodiments.