Agile Robots and NVIDIA Unveil Agile ONE SN Humanoid Robot for Research, Set for Early 2027 Launch | Linear Portfolio

y'x

At NVIDIA's recent GTC conference, Agile Robots and NVIDIA jointly unveiled the Agile ONE SN, a humanoid robot designed for research.

Built on the Agile ONE S humanoid platform, the Agile ONE SN integrates a compact body, tactile five-finger dexterous hands, multi-view perception, and onboard AI computing into a single system, providing the hardware foundation for research in motion control, environmental perception, and dexterous manipulation.

Linear Capital was the lead investor in Agile Robots' angel round and has continued to invest in subsequent rounds, supporting the company all the way to becoming a leading unicorn in the global robotics sector.

On October 21, at the NVIDIA GTC conference, Agile Robots and NVIDIA jointly launched the Agile ONE SN, a humanoid robot for research applications. Built on the NVIDIA Isaac™ GR00T development platform, it comes pre-configured with a complete software stack out of the box, providing researchers and developers worldwide with a "ready-to-use" full-stack humanoid robot platform.

The advancement of general-purpose humanoid robots is placing greater demands on R&D tools and infrastructure. Robots need to understand tasks, perceive changes, and coordinate their bodies and hands to complete operations in real-world environments. Research teams, meanwhile, need to connect data collection, simulation, training, and deployment to continuously validate and improve robot capabilities.

However, these stages often rely on disparate tools. Hardware adaptation, software configuration, and workflow integration consume significant development time, and complicate the transition from simulation to physical robot testing. The Agile ONE SN unifies these stages into a single platform, allowing researchers to begin experiments faster and focus more energy on robot skills themselves.

Dr. Zhaopeng Chen, founder and CEO of Agile Robots, said: "We are committed to rapidly bringing embodied intelligence products from the lab to real-world applications. The Agile ONE SN provides global R&D labs with a complete, ready-to-use humanoid robot platform. Working with NVIDIA, we are helping them turn ideas into robot skills faster, and paving a clear path for those skills to reach practical deployment."

Designed around the needs of humanoid robot R&D, the Agile ONE SN integrates a compact body, tactile five-finger dexterous hands, multi-view perception, and onboard AI computing based on the Agile ONE S platform, providing the hardware foundation for research in motion control, environmental perception, and dexterous manipulation.

• Agile ONE SN humanoid body: 1.4 meters tall, 43 kg
• 31 degrees of freedom across the body for human-like movement
• Onboard NVIDIA Jetson Thor™ computing platform for real-time sensor processing and robot inference
• Multi-view sensing: head-mounted RGB-D camera, wrist cameras, and 3D LiDAR, combined with an inertial measurement unit
• Whole-body control: joint torque up to 140 N·m, single-arm payload up to 3 kg
• Wi-Fi 7 and Bluetooth connectivity, with emergency stop button and other safety features

On the software side, the Agile ONE SN comes pre-installed with NVIDIA Isaac GR00T, including the GR00T 1.7 open-source vision-language-action (VLA) model, GR00T whole-body control, and open-source end-to-end reference workflows covering demonstration data collection, data generation, simulation, training, evaluation, deployment, and real-world testing.

This workflow provides research teams with a continuous development path: starting from task demonstration collection, moving to training and evaluation in simulation, then deploying robot skills to the physical platform for validation, and iterating based on test results. Data, models, and hardware experiments are thus more tightly connected.

The platform also provides URDF and USD assets, allowing researchers to import the robot directly into NVIDIA Isaac Sim and Isaac Lab for simulation and robot learning research. The integration of tools and workflows in the Agile ONE SN can reduce setup time from weeks to days, helping teams begin real-world experiments sooner.

Deepu Talla, Vice President of Robotics and Edge AI at NVIDIA, said: "Advancing humanoid robot R&D requires convenient access to a complete, open hardware and software platform. The Agile ONE SN integrates NVIDIA Isaac GR00T and Jetson Thor into a single system, unifying demonstration data collection, simulation, training, validation, and deployment. This reduces integration work and accelerates the development of humanoid robot skills for real-world applications."

Openness and modularity are also core design principles of the Agile ONE SN. Research teams can use the complete reference workflow or select components based on their research direction and integrate them into their own technology stack, gaining platform support while retaining development flexibility and maintaining control over their own data. This also facilitates sharing research methods and comparing experimental results across different labs.

In terms of safety capabilities, Agile Robots is exploring integration with the NVIDIA Halos safety platform, advancing work around design, validation, and software to build the technical foundation for humanoid robots to enter real-world application environments.

The launch of the Agile ONE SN marks another important step in Agile Robots' ongoing effort to connect cutting-edge research with practical applications. Drawing on long-accumulated experience in research collaboration, the platform is aimed at global R&D institutions to support frontier humanoid robot research, and has already attracted interest from institutions including Stanford University and ETH Zurich.

At this GTC conference in Berlin, NVIDIA also announced the expansion of the Isaac open robot development platform, introducing new robot learning tools, datasets, and workflows. The Agile ONE SN combines Agile Robots' hardware and system integration capabilities with NVIDIA's models, simulation, and computing platforms to provide developers with a ready-to-research hardware and software system. NVIDIA founder and CEO Jensen Huang said at GTC: "General-purpose robots are advancing at a remarkable pace — the 'ChatGPT moment' for robots is fast approaching."

From a single skill training session in the lab to continuous validation against real-world tasks, humanoid robots' path to industry requires robust R&D tools and a vehicle that connects research with application. With the launch of the Agile ONE SN, Agile Robots will work with NVIDIA and developers worldwide to transform more research ideas into robot capabilities that can be trained, validated, and deployed.

The Agile ONE S/SN is planned for release in early 2027. It will connect with the full-size humanoid robot Agile ONE for industrial applications, providing new R&D support for embodied intelligence's journey from the lab to industry.