Isaac ROS Grasping#

Overview#

GraspGenX generates ranked grasp candidates from segmented point clouds using native C++/CUDA preprocessing and TensorRT inference. Its standalone quickstart builds CUDA point clouds from registered depth and object masks, runs inference, and displays the resulting gripper poses in RViz.

Packages#

Requirements#

Use the Isaac ROS source environment with CUDA-backed rosidl::Buffer support, compatible middleware, and a CUDA GPU. Build the TensorRT engine and runtime plugin for the same GPU and CUDA/TensorRT environment. Refer to the model setup requirements.

Quickstart#

Supported Platforms#

This package is designed and tested to be compatible with ROS 2 Lyrical running on Jetson.

Platform

Hardware

Software

Storage

Notes

Jetson

Jetson Thor (T5000 and T4000) and Jetson Orin

JetPack 7.2

128+ GB NVMe SSD

For best performance, ensure that power settings are configured appropriately for your platform: Jetson Thor power settings or Jetson Orin power settings.

Isaac ROS Environment#

To simplify development, we strongly recommend leveraging the Isaac ROS CLI by following these steps. This streamlines your development environment setup with the correct versions of dependencies on all supported platforms.

Note

All Isaac ROS Quickstarts, tutorials, and examples have been designed with the Isaac ROS CLI-managed environment as a prerequisite.

Customize your Dev Environment#

To customize your development environment, refer to this guide.

Updates#

Date

Changes

2026-11-16

Initial release of GraspGenX for single-object and batch grasp generation