datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
BODex_lfs
BODex
A GPU-based efficient pipeline for dexterous grasp synthesis built on cuRobo, proposed in [ICRA 2025] BODex: Scalable and Efficient Robotic Dexterous Grasp Synthesis Using Bilevel Optimization.
Project page ｜ Paper ｜ Dataset ｜ Benchmark code ｜ Learning code
Introduction
Main Features
Grasp Synthesis: Generate force-closure grasps for floating dexterous hands, such as the Shadow, Allegro, and Leap Hand.
Trajectory Planning: Plan collision-free… See the full description on the dataset page: https://huggingface.co/datasets/JiayiChenPKU/BODex_lfs.BODex[ICRA 2025] BODex: Scalable and Efficient Robotic Dexterous Grasp Synthesis Using Bilevel Optimization.
Project page | Paper | Grasp synthesis code | Grasp validation code | Object pre-processing code | Learning code
Dataset structure:
object_assets
|- DGN_2k_processed.tar.gz # Our pre-proposed meshes and training splits.
|- DGN_2k_vision.tar.gz # The single-view point clouds. Only used for training networks.
|_ DGN_obj_raw.zip # The raw object meshes before pre-processing. Not… See the full description on the dataset page: https://huggingface.co/datasets/JiayiChenPKU/BODex.SimVLA-BODex
SimVLA-BODex
Preprocessed object meshes packaged for use with the
SimVLA project. The contents are a
re-distribution of meshes generated by the
BODex preprocessing pipeline.
Contents
The dataset ships as 3 tarballs (~2.1 GB compressed, ~6.3 GB extracted).
Tarball
Compressed
Contents
graspdata_final.tar.gz
~1.6 GB
Grasp pose data
processed_data.tar.gz
~0.5 GB
Full object set, 2,339 dirs
use_data.tar.gz
~28 MB
Curated 129-object kitchen subset — what… See the full description on the dataset page: https://huggingface.co/datasets/exaFLOPs09/SimVLA-BODex.