Agile Robots unveils Nvidia-powered humanoid built to learn new skills

Agile Robots has introduced Agile ONE SN, a humanoid platform that combines a robot, onboard computing, and Nvidia’s AI training tools.

The system’s main feature is that these components come integrated, giving researchers a ready-made starting point for developing robot skills. The companies say this could cut experiment preparation from weeks to days.

Developers can use Nvidia’s Isaac GR00T platform to collect training data, adapt AI models, test behavior in simulation, and deploy trained skills on the physical robot. A Jetson Thor computer runs the AI onboard and supports real-time movement control, connecting what the robot perceives with how it acts.

Agile ONE SN stands 1.4 meters tall, weighs 40 kilograms, and has 31 degrees of freedom for moving its body and limbs. Each arm can carry up to 3 kilograms. Cameras in its head and wrists, lidar, and motion sensors help it perceive its surroundings.

Its two five-finger WUJI Hand 2 hands each have 20 degrees of freedom, tactile sensors, and force control operating at 1,000 updates per second. These features are intended to support more precise handling of objects. The companies also plan to integrate Nvidia’s Halos safety technology.

Agile Robots CEO Zhaopeng Chen described the goal as an adaptable machine that learns multiple skills and works in environments built for people. Nvidia CEO Jensen Huang said “the ChatGPT moment for robotics is fast approaching,” framing the announcement as part of a broader push toward general-purpose robots.

Agile ONE SN is scheduled to become available in early 2027. The announcement identifies Stanford University and ETH Zurich among institutions that have expressed interest in the platform.