Skip to content

Dataset · unitreerobotics

Telemetry and video of cushion grasping by Unitree G1 humanoid

unitreerobotics/G1_WBT_SMPL_Inspire_Pick_Up_Cushions

From 10k to 100k records with video, joint poses, and finger poses of the Unitree G1 robot during cushion pickup.

What's inside

The official dataset from Unitree Robotics contains 10k to 100k steps in the form of video, tabular data, and time series. The observations record 36 parameters of the robot's current configuration (body position and orientation in space plus angles of 29 joints), as well as the states of the Inspire Hand manipulators' fingers.

What you can do with it

  • Training control models for a humanoid robot's hands and body
  • Training grasping policies for soft and deformable objects (cushions)
Tasks
Robotics
Size
10K–100K rows
Downloads this month
1.2K
Likes
2

How to load it

from datasets import load_dataset

ds = load_dataset("unitreerobotics/G1_WBT_SMPL_Inspire_Pick_Up_Cushions")