DefGraspSim: Simulation-based grasping of 3D deformable objects


Isabella Huang, Yashraj Narang, Clemens Eppner, Balakumar Sundaralingam, Miles Macklin, Tucker Hermans, Dieter Fox

Overview

This website contains the following supplementary information for our submission:

  1. Additional videos of simulated and real experiments

  2. The codebase developed for our simulations, which can be used to simulate and evaluate grasps on arbitrary 3D deformable objects

  3. The datasets containing measured grasp features and performance metrics on object primitives and complex objects

  4. Interactive visualizations of grasp results on object primitives

Additional videos

Simulated grasp tests (left), and real-world validation experiments (right)

rss_p1.mp4
rss_p2.mp4

Isaac Gym and Code Release

The preview release of Isaac Gym is now live: https://developer.nvidia.com/isaac-gym

Access the code used in these experiments: https://github.com/NVlabs/defgraspsim

A video tutorial on how to get started with the code is here: https://youtu.be/Caj0AtsKKVI

Datasets

The live dataset on object primitives and complex objects will be continually updated. Click here for details on the contents of each file.

Interactive Visualizations

Click here to interactively visualize grasps and performance metrics on 9 different object primitives.