NVIDIA Launches Isaac ROS 5.0 to Advance Agentic and Open Source Robotics
NVIDIA has released Isaac ROS 5.0 at ROSCon, introducing AI agent capabilities, GPU-accelerated physical AI libraries, and broad support for the Jetson platform.
New AI Agent Capabilities in Isaac ROS 5.0
NVIDIA has released Isaac ROS 5.0, a collection of GPU-accelerated packages built on the Robot Operating System (ROS). Unveiled at the ROSCon conference in Toronto, Canada, the update introduces new agentic workflows designed to help developers and AI agents build, customize, and deploy robotics applications faster.
The latest version brings support for ROS Lyrical and Ubuntu 24.04, providing developers with updated tools while leveraging NVIDIA accelerated computing. Additionally, NVIDIA worked with the Open Source Robotics Alliance to contribute a standard data-handling interface to ROS Lyrical, helping software operate efficiently across diverse computing hardware including GPUs.
Expanding Physical AI and Workflows
Isaac ROS 5.0 introduces new reusable skills for setup and manipulation that both human developers and AI agents can utilize. Agent-ready documentation is also included to help AI agents interpret tools and workflows, converting developer intent into functional applications with greater efficiency.
The release features several upgraded perception and manipulation capabilities, including a FoundationStereo fine-tuning skill to adapt stereo perception models to specific environments. Furthermore, an agent-ready inference library for the FoundationPose object pose estimation model allows robots to track object positions up to 5.5 times faster, alongside a standalone pick-and-place workflow skill.
Deployment Across the Jetson Ecosystem
Designed to scale from entry-level hardware like the NVIDIA Jetson Orin Nano to high-performance Jetson Thor devices, Isaac ROS 5.0 bridges development and edge deployment. Several robotics companies, including Mentee Robotics, Universal Robots, and ROBOTIS, are integrating the new software packages into their physical AI and humanoid platforms.
The software is available now as a free and open-source release, with resources hosted on GitHub for developers looking to incorporate accelerated perception, mapping, and motion planning into their robotic stacks.
Source
Official announcement or documentation
Some links on this page may be affiliate links. If you buy through them we may earn a commission at no extra cost to you. See our affiliate disclosure.
