News from IT on October 21st: On October 21st, local time, NVIDIA announced at the GTC Berlin conference that it would expand the Isaac Open Robotics Development Platform. New features include a humanoid robot reference platform, a runtime framework, simulation assets, and datasets, as well as updated robotics learning tools. These improvements cover the entire development process, from training data generation and simulation training to the deployment of physical robots.
In terms of hardware, NVIDIA, in collaboration with Singularity Robotics ( Agile Robots ), has released a humanoid robot reference platform called Agile ONE SN, which is scheduled to be launched at the beginning of 2027.
The platform is pre-configured with NVIDIA Isaac GR00T and is equipped with the Jetson Thor computing platform, mainly aimed at robot research and development. By integrating simulation, training, and deployment tools, Agile Robots indicates that this platform can reduce the preliminary configuration time for related environments from several weeks to just a few days.
NVIDIA founder and CEO Jensen Huang stated that general-purpose robotics technology is developing rapidly, and the "ChatGPT moment" in the field of robotics is accelerating its arrival. NVIDIA claims that this update is aimed at providing developers with models, data, simulation tools, and computing systems to help them apply general-purpose robotics technology to real industrial scenarios.
SIRI robot Agile ONE SN uses a Agile ONE S humanoid robot body, with a height of 1.4 meters and a weight of 40 kilograms, featuring 31 degrees of freedom, and a single-arm load capacity of 3 kilograms. It is equipped with two WUJI Hand 2-finger dexterous hands, each with 20 degrees of freedom, supporting 1kHz force control and tactile perception. It should be noted that according to Reuters, the weight of this robot is approximately 34 kilograms, which differs from the 40 kilograms officially announced by NVIDIA; however, the current information is not sufficient to determine the cause of this discrepancy.
In terms of computing, Agile ONE SN is equipped with NVIDIA Jetson Thor, providing 2070 TFLOPS of FP4 computing power and 128GB of unified memory. The sensors include cameras on the head and wrist, lidar, and an inertial measurement unit, which are used to perceive the surrounding environment and support interaction between the robot and the external world.
In terms of security, the platform plans to integrate the robot-side security protocol with the NVIDIA Halos full-stack security system, although this integration is still pending completion. NVIDIA stated that research institutions such as Stanford University and the Swiss Federal Institute of Technology Zurich have shown interest in Agile ONE SN.
Reuters also reported that Agile Robots plans to collaborate with NVIDIA in 2027 to produce up to 2,000 humanoid robots, targeting customers such as university and corporate research laboratories.
In terms of software, NVIDIA Isaac Lab 3.0 has been officially released. As an open-source robotics learning framework, it supports multiple GPU training workflows and can utilize physical engines such as Newton and PhysX. This framework is designed to assist developers in training and optimizing robot strategies, covering a range of computing environments from desktop workstations to data centers.
IT noticed that NVIDIA has also launched an open-source runtime framework, Isaac OS. This framework is built on Holoscan and is used to connect AI models, sensors, control systems, and robot hardware. It supports migration from simulation environments to physical robots and provides deterministic real-time operation capabilities.
For DGX Station, NVIDIA released three sets of open-source reference workflows. The first set optimizes the GR00T 1.7 model through reinforcement learning; the second set uses Isaac Lab to train dexterous operation skills; the third set utilizes ChatGPT Astra to perform post-training on the Cosmos world model in order to generate synthetic training data. These workflows are designed to help developers integrate robot learning, simulation, and data generation processes.
In terms of data, SimReady Central Library provides over 2000 robot and factory assets that can be used for simulation, covering a variety of robotic equipment and industrial scenario resources. The Human 1M dataset contains various human activities, synchronized three-camera videos, and 3D hand tracking data, which can be used for robot learning and motion understanding. Both resources are provided through Hugging Face, and the full version of the Human 1M dataset is only available to Cosmos Coalition partners.
NVIDIA also introduced several companies that are using its robot development tools. ABB Robotics has integrated the Omniverse library into RobotStudio HyperReality for simulation and virtual debugging before robot system installation; NEURA Robotics uses Isaac Sim and Isaac Lab to generate synthetic training data and train robot skills in a simulated environment. NVIDIA stated that companies such as KUKA, Humanoid, Sereact, and Universal Robots are also adopting its open robot platform-related technologies.
