The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
Gripper Descriptions
Real-world gripper URDF descriptions, meshes, and configuration files for GraspGenX.
Supported Grippers
![]() abb_yumiparallel_2f |
![]() arx_x5revolute_2f |
![]() barrett_handrevolute_3f |
![]() bd_spotrevolute_2f |
![]() dh_ag95parallel_2f |
![]() ezgripperrevolute_2f |
![]() fetch_robotparallel_2f |
![]() franka_pandaparallel_2f |
![]() franka_umiparallel_2f |
![]() galaxea_g1revolute_2f |
![]() inspire_handrevolute_3f |
![]() onrobot_RG2revolute_2f |
![]() onrobot_RG6revolute_2f |
![]() piper_handrevolute_2f |
![]() robotiq_2f_140revolute_2f |
![]() robotiq_2f_85revolute_2f |
![]() robotiq_3frevolute_3f |
![]() robotiq_handeparallel_2f |
![]() sawyer_handparallel_2f |
![]() schunk_wsg50parallel_2f |
![]() sharpa_waverevolute_3f |
![]() surge_handrevolute_3f |
![]() tesollo_delto2frevolute_2f |
![]() unitree_g1revolute_2f |
![]() wuji_handrevolute_3f |
![]() xarm_handrevolute_2f |
Installation
pip install -e .
Or directly from the repository:
pip install -e git+ssh://git@gitlab-master.nvidia.com:12051/rays2torques/graspgen/gripper_descriptions.git#egg=gripper_descriptions
If the git lfs clone failed, download the assets from: https://drive.google.com/file/d/1_GlOZjy5treLxb3Q30V3VdMlssOIWdy0/view?usp=sharing and put it under gripper_descriptions/assets/ .
Usage
import gripper_descriptions
# Get path to all gripper assets
assets_path = gripper_descriptions.get_assets_path()
# Get path to a specific gripper
gripper_path = gripper_descriptions.get_gripper_path("robotiq_2f_85")
# List all available grippers
print(gripper_descriptions.list_grippers())
print(gripper_descriptions.AVAILABLE_GRIPPERS)
With GraspGenX
Once installed, GraspGenX automatically discovers gripper assets from this package:
from graspgenx.x_grippers import get_gripper_info
import gripper_descriptions
gripper = get_gripper_info(gripper_descriptions.get_assets_path(), "robotiq_2f_85")
Visualization
# Visualize a single gripper
python -m gripper_descriptions.scripts.vis_gripper --gripper robotiq_2f_85
# Animate open/close
python -m gripper_descriptions.scripts.vis_gripper --gripper robotiq_2f_85 --animate
# Show all grippers in a grid
python -m gripper_descriptions.scripts.vis_all_grippers
# List available grippers
python -m gripper_descriptions.scripts.vis_gripper --list-grippers
Asset Structure
Each gripper directory contains:
| File | Description |
|---|---|
gripper.urdf |
URDF with mesh paths relative to meshes/ |
meshes/ |
STL/OBJ mesh files referenced by the URDF |
config.json |
Joint configs (open/close), fingertip, sweep volume, type |
vis_mesh.obj |
Merged visual mesh (open configuration) |
coll_mesh.obj |
Merged collision mesh (open configuration) |
points.json |
Surface point clouds (open/close, 10500 pts each) |
selected_pts.json |
Inner-surface point clouds (open/mid/close) |
tsdf.npy |
Truncated signed distance fields |
proc_gripper_only_pointnet_vae_repr.json |
PointNet VAE embedding (64-D) |
- Downloads last month
- 112

























