NVIDIA and Agile Robots Launch Agile ONE SN Humanoid Robot Platform for 2027 Deployment

IT Home On October 21, news reported that NVIDIA announced during the GTC Berlin conference on October 21 the expansion of the Isaac open robot development platform, adding a humanoid robot reference platform, runtime framework, simulation assets, and datasets, and updating robot learning tools to cover the development process from training data generation, simulation training to physical robot deployment.

In terms of hardware, NVIDIA collaborated with Agile Robots to release the humanoid robot reference platform Agile ONE SN, planned to be launched in early 2027.

The platform is pre-configured with NVIDIA Isaac GR00T and equipped with the Jetson Thor computing platform, primarily targeting robot research and development. By integrating simulation, training, and deployment tools, Agile Robots stated that the platform can reduce the initial setup time for relevant environments from weeks to days.

NVIDIA founder and CEO Huang Renxun stated that general-purpose robot technology is rapidly evolving, and the 'ChatGPT moment' in the robotics field is accelerating. NVIDIA stated that this update aims to provide developers with models, data, simulation tools, and computing systems to help them apply general-purpose robot technology to real-world industrial scenarios.

The Agile ONE SN from Sighan Robotics adopts the Agile ONE S humanoid robot body, standing at 1.4 meters tall and weighing 40 kilograms, with 31 degrees of freedom and a single arm payload of 3 kilograms. It is equipped with two WUJI Hand 2 five-finger dexterous hands, each hand having 20 degrees of freedom, supporting 1kHz force control and tactile perception. It is noted that Reuters reported the robot's weight to be approximately 34 kilograms, differing from NVIDIA's official data of 40 kilograms. Existing information is insufficient to confirm the reason for the discrepancy.

In terms of computing, Agile ONE SN is equipped with NVIDIA Jetson Thor, providing 2070 TFLOPS of FP4 computing power and 128GB unified memory. Sensors include head and wrist cameras, LiDAR, and inertial measurement units, used for environmental perception and to support interaction between the robot and the external environment.

In terms of safety, the platform plans to integrate robot-end safety protocols with NVIDIA Halos full-stack security system, although this integration is still pending. NVIDIA stated that research institutions such as Stanford University and ETH Zurich have expressed interest in Agile ONE SN.

Reuters also reported that Agile Robots plans to produce up to 2,000 humanoid robots in collaboration with NVIDIA by 2027, targeting customers such as university and enterprise research laboratories.

In terms of software, NVIDIA Isaac Lab 3.0 has been officially released. As an open-source robot learning framework, it supports multi-GPU training workflows and can use physics engines such as Newton and PhysX. The framework aims to help developers train and optimize robot strategies, covering different computing environments from desktop workstations to data centers.

IT Home noticed that NVIDIA also released an open-source runtime framework Isaac OS. This framework is built on Holoscan, used to connect AI models, sensors, control systems, and robot hardware, supporting migration from simulation environments to physical robots, and providing deterministic real-time runtime capabilities.

Regarding DGX Station, NVIDIA also released three open-source reference workflows. The first optimizes the GR00T 1.7 model through reinforcement learning; the second trains dexterous manipulation skills using Isaac Lab; the third leverages ChatGPT Astra to post-train the Cosmos world model, generating synthetic training data. These workflows aim to help developers integrate robot learning, simulation, and data generation processes.

In terms of data, SimReady Central Library provides over 2,000 items of robot and factory assets for simulation, covering various robot devices and industrial scenario resources. The Human 1M dataset includes various human activities, synchronized three-view video, and 3D hand tracking data, which can be used for robot learning and action understanding. Both resources are provided through Hugging Face, with the complete version of the Human 1M dataset only available to Cosmos Coalition partners.

NVIDIA also introduced several companies adopting its robot development tools. ABB Robotics integrated the Omniverse library into RobotStudio HyperReality for simulation and virtual debugging before robot system installation; NEURA Robotics used Isaac Sim and Isaac Lab to generate synthetic training data and trained robot skills in simulation environments. NVIDIA stated that companies such as KUKA, Humanoid, Sereact, and Universal Robots are also adopting related technologies from its open robot platform.