NVIDIA Showcases Robotics Innovations at CoRL 2025 Conference
NVIDIA is seizing the spotlight at the Conference on Robot Learning (CoRL) in Korea to unveil its latest advancements in robotics learning. The company’s showcase focuses on a range of new models, powerful computer systems, and simulation tools designed to bridge the gap between theoretical robotic research and real-world applications.
Central to NVIDIA’s announcements are the challenges of mobility and manipulation, which are among the most intricate obstacles in robotics. By integrating AI-driven reasoning, cutting-edge physics simulation, and specialized computational platforms, NVIDIA is striving to expedite the entire robotic development process, from initial research and simulation to training and deployment in practical environments.
NVIDIA has introduced two open-source foundation models named DreamGen and Nerd, tailored for robotic learning. By making these models publicly available, NVIDIA is underlining its dedication to communal robotics research, simplifying access for research labs and startups, and ensuring consistency across academic and industrial domains. These models seamlessly integrate with the Isaac robotics frameworks, streamlining experimentation and implementation processes.
In addition, NVIDIA has unveiled three new computing systems optimized for different stages of robotics development: Omniverse with Cosmos on RTX Pro (for Simulation), NVIDIA DGX (for Training), and NVIDIA Jetson AGX Thor (for Deployment). The Newton Physics Engine, a collaboration with Disney Research and Google DeepMind, was debuted at GTC 2025. NVIDIA also showcased the advancement of its Groot model family for robotic reasoning, progressing from Groot N1.5 to Groot N1.6, which are fine-tuned for human-like planning and task breakdown.
Highlighting the real-time reasoning capabilities in robotics facilitated by the new Blackwell GPU architecture, NVIDIA’s exhaustive robotics initiative not only introduces a wide array of tools but also demonstrates the deeply intertwined ecosystem of the advancements. By combining open research models, accurate physics simulations, and deployable computing resources, NVIDIA is positioned as a pivotal driver in propelling robotic autonomy forward.
About the Author
