NVIDIA expands Isaac platform for general-purpose robotics

New layers for the Isaac platform

At GTC Berlin, NVIDIA announced an expansion of its Isaac open robotics development platform. The package adds a humanoid reference platform, datasets, a new runtime and updates to its robot learning framework. The goal is to give developers an integrated path from data to deployment.

Agile ONE SN

Announced by Agile Robots, Agile ONE SN is a full-stack humanoid reference platform preconfigured with Isaac GR00T and powered by Jetson Thor. It stands 1.40 meters tall and weighs 40 kilograms, with a 3-kilogram payload per arm and 31 degrees of freedom. Its dual WUJI Hand 2 hands are five-fingered, each with 20 degrees of freedom and tactile sensing.

Jetson Thor delivers 2,070 FP4 teraflops of compute and 128 GB of unified memory for onboard reasoning and real-time control. The platform also includes head and wrist cameras, lidar and inertial measurement units. NVIDIA says Agile ONE SN is scheduled to be available in early 2027, with interest from institutions such as Stanford and ETH Zurich.

Software and data pipeline

NVIDIA added the Isaac OS runtime and Isaac Lab 3.0 for robot learning. Open source reference workflows that combine ChatGPT Astra, Isaac and Cosmos on DGX Station aim to accelerate robot data generation, training and evaluation. Integration of the NVIDIA Halos safety stack into Agile ONE SN is planned.

European ecosystem

European robotics companies including ABB Robotics, KUKA, Universal Robots, NEURA Robotics, Humanoid, Hexagon Robotics, Flexion and Sereact are building autonomous systems on NVIDIA's open platform. Jensen Huang said the "ChatGPT moment" for general-purpose robotics is fast approaching.