Insider Brief
- Nvidia released Isaac ROS 5.0, adding AI-agent tools, expanded ROS support and new perception and manipulation capabilities to its open-source robotics software.
- The update adds support for ROS Lyrical and Ubuntu 24.04, plus reusable skills for setup, manipulation, FoundationStereo fine-tuning, FoundationPose and pick-and-place workflows.
- Nvidia said Isaac ROS 5.0 is already being integrated by companies including RealSense, Intrinsic, Magna, Mentee Robotics, Universal Robots and Robotis, and is available now through GitHub.
Nvidia has released Isaac ROS 5.0, an update to its open-source robotics software that adds tools for AI agents, expanded ROS support and new capabilities for perception and manipulation.
According to Nvidia, Isaac ROS is a collection of GPU-accelerated software packages built on the Robot Operating System, or ROS. The latest version is designed to help developers use AI agents alongside conventional robotics tools to set up, customize and deploy applications.
Isaac ROS 5.0 adds support for ROS Lyrical and Ubuntu 24.04. Nvidia also worked with the Open Source Robotics Alliance on a standard interface for moving data efficiently across computing hardware, including GPUs. The interface is available to the broader ROS community, with CUDA serving as one implementation for GPU acceleration.
The release also introduces reusable “skills” that developers and AI agents can use for robotics development tasks. Key additions include:
- Setup and manipulation skills: Reusable workflows for configuring and building robotics applications.
- FoundationStereo fine-tuning: Allows an AI agent to help adapt a stereo-perception model to specific cameras, environments and applications.
- FoundationPose: Nvidia said its object pose-estimation and tracking model can now detect and track the position and orientation of objects up to 5.5 times faster.
- Pick-and-place: A standalone skill that combines object detection, depth estimation and pose information for manipulation tasks.
How Isaac ROS is Used
The software is designed to run across Nvidia’s Jetson edge-computing hardware, from Jetson Orin Nano through Jetson Thor. Nvidia said this allows developers to use the same platform as applications move from development into robots.
Several robotics companies are already integrating Isaac ROS into their development systems, according to Nvidia. RealSense-sponsored AgenticROS connects Isaac ROS with Nvidia Nemotron models and NemoClaw blueprints so AI agents can interact with ROS-based robots. RealSense is also optimizing its D585 Pro stereo depth camera and software development kit for Isaac ROS and Nvidia Jetson Thor.
Intrinsic’s newly released Open Machine Tending Solution uses Nvidia FoundationPose for object registration, tracking and pose estimation. Seeed Studio is using Isaac ROS with its reBot Arm on Jetson Thor, while Magna is pairing Isaac ROS with Isaac Sim for perception, synchronized data collection and testing. Flexiv is integrating the software into its Rizon 4 adaptive robot, and Ekumen is using Isaac ROS packages for docking, obstacle detection, visual localization and motion planning.
Mentee Robotics is using Isaac ROS for perception and AI processing in its MenteeBot humanoid, while Universal Robots has incorporated it into its AI Accelerator software development kit. Robotis is integrating it into its AI Worker robot for vision-guided manipulation, and FieldAI and Noble Machines are using Isaac ROS on Jetson hardware for on-robot AI processing and industrial robotics development.
Isaac ROS 5.0, which was announced at ROSCon 2026 in Toronto, is available through GitHub.
Image credit: Nvidia
