Katie Washabaugh

NVIDIA released Isaac ROS 5.0 at ROSCon in Toronto, introducing agentic workflows and new platform support for robotics development. The update adds GPU-accelerated tools, AI agent-ready documentation, and new skills like FoundationStereo fine-tuning and FoundationPose inference that enable faster perception and object tracking. The release targets the 1.3 million ROS users and includes support for ROS Lyrical and Ubuntu 24.04, positioning NVIDIA's accelerated computing as a standard for high-performance robotics applications.

NVIDIA Research presented eight papers at ICRA demonstrating how simulation-to-real transfer is enabling robots to operate reliably in dynamic, unpredictable environments. The work spans multi-arm coordination, cross-embodiment navigation policies, adaptive grasping, and deformable object manipulation, all trained in simulation without real-world robot data. The research shows measurable improvements: 3x speedup in multi-arm planning, 4.5x better navigation success rates, and 75% grasping success on real robots versus 41% baseline.

NVIDIA has released new open models and frameworks designed to streamline the development of production robots by integrating simulation, robot learning, and embedded compute into unified cloud-to-robot workflows. The tools aim to reduce friction in moving AI systems from simulated environments to real-world robotic hardware. This represents a shift toward making robot development more accessible and faster by consolidating previously fragmented tooling and infrastructure layers.