NVIDIA Unveils Isaac GR00T N1.5 and GR00T-Dreams to Propel Humanoid Robotics at Computex 2025.

Advancing Humanoid Intelligence with Isaac GR00T N1.5:

NVIDIA has released Isaac GR00T N1.5, the first major update to its open, customizable foundation model designed for humanoid robot reasoning and skills. This model enhances a robot's ability to adapt to new environments, recognize objects through user instructions, and perform complex tasks such as sorting and object placement. The GR00T N1.5 model can be deployed on the upcoming NVIDIA Jetson Thor robot computer, expected later this year.

GR00T-Dreams: Generating Synthetic Motion Data:

To address the challenges of acquiring extensive real-world data for training robots, NVIDIA introduced the GR00T-Dreams blueprint. This tool leverages the Cosmos platform to generate vast amounts of synthetic motion data, known as neural trajectories, from minimal human demonstrations. Developers can input a single image, and GR00T-Dreams will produce videos of robots performing new tasks in varied environments. These videos are then analyzed to extract action tokens, which are used to train robots in executing these tasks.

The GR00T-Dreams blueprint complements the previously released GR00T-Mimic, which augments existing data using NVIDIA Omniverse and Cosmos platforms. While GR00T-Mimic focuses on enhancing existing datasets, GR00T-Dreams is designed to generate entirely new data, thereby closing the data gap in robot training.

Accelerating Development with Simulation Frameworks:

NVIDIA also unveiled several simulation technologies to expedite robot training:

Cosmos Reason:

A new world foundation model that uses chain-of-thought reasoning to curate high-quality synthetic data for physical AI model training.

An enhanced model for high-quality world generation with reduced hallucination, used in GR00T-Dreams.

Isaac Sim 5.0: A simulation and synthetic data generation framework, soon to be available on GitHub, providing developers with tools to test and validate robot behaviors.

Isaac Lab 2.2:

An open-source robot learning framework supporting new evaluation environments for testing GR00T N models.

These tools collectively aim to reduce the time and resources required for developing and deploying humanoid robots.

Industry Adoption and Applications:

Several robotics developers are integrating NVIDIA's Isaac platform technologies into their workflows:

Agility Robotics, Boston Dynamics, Fourier, Foxlink, Galbot, Mentee Robotics, NEURA Robotics, General Robotics, Skild AI, and XPENG Robotics are adopting these technologies to advance humanoid robot development and deployment.

AeiRobot utilizes the GR00T N model to enable its ALICE4 robot to understand natural language instructions and execute complex pick-and-place workflows in industrial settings.

Foxlink Group is enhancing industrial robot manipulator flexibility and efficiency using the model.

NVIDIA Newsroom:

Lightwheel employs the model to validate synthetic data for faster humanoid robot deployment in factories.

NEURA Robotics is evaluating the model to accelerate the development of household automation systems.

Empowering Development with NVIDIA Blackwell Systems:

To support the computational demands of training and deploying advanced AI models, NVIDIA introduced the RTX PRO 6000 workstations and servers, powered by the Blackwell architecture. These systems offer a unified architecture for robot development workloads, including training, synthetic data generation, robot learning, and simulation.

Global systems manufacturers such as Cisco, Dell Technologies, Hewlett-Packard Enterprise, Lenovo, and Supermicro have announced RTX PRO 6000 Blackwell-powered servers. Additionally, Dell Technologies, HPI, and Lenovo have introduced NVIDIA RTX PRO 6000 Blackwell-powered workstations.

For large-scale training or data-generation workloads, developers can utilize Blackwell systems like the GB200 NVL72, available through NVIDIA DGX Cloud and NVIDIA Cloud Partners, to achieve up to 18 times greater performance for data processing. These models can be deployed to the upcoming NVIDIA Jetson AGX Thor platform to accelerate on-robot inference and runtime.

NVIDIA's latest offerings at Computex 2025 underscore its commitment to advancing physical AI and humanoid robotics. By providing comprehensive tools and platforms for data generation, simulation, and deployment, NVIDIA aims to accelerate the development and integration of intelligent robots across various industries.

Source:https://www.therobotreport.com/nvidia-cloud-robot-computing-platforms-physical-ai-humanoid-development/

This is non-financial/medical advice and made using AI so could be wrong.