Germany's Agile Robots launches humanoid research platform for 2027

A unified research stack

Germany-based Agile Robots recently introduced the Agile ONE SN, a compact humanoid platform designed specifically for research and development. As reported by AI Insider, the robot is intended to help developers build, train, and test robot skills within a single integrated system.

Currently, many research labs must connect separate, disconnected systems to handle different parts of the robotics pipeline. This includes tools for data collection, digital simulation, model training, and final deployment on physical hardware. The Agile ONE SN aims to consolidate these steps into one workflow to move experiments from the lab to the real world more quickly.

Hardware and sensing specs

The Agile ONE SN stands 1.4 meters tall and weighs 43 kilograms. It is equipped with 71 total degrees of freedom, which refers to the number of independent joints that can move. This includes 31 joints in the body and 40 joints across two five-finger Wuji tactile hands.

The robot's sensing suite is designed for complex manipulation and navigation. It features 3D lidar for spatial awareness, RGB-D cameras on the head, and specialized wrist cameras for close-range tasks. Each arm can handle a payload of up to 3 kilograms, while the joints provide torque of up to 140 Nm.

For computing power, the robot utilizes Nvidia Jetson Thor. This hardware handles real-time sensor processing and robot inference, which is the process of the AI making decisions based on its inputs.

Software and simulation

The platform ships with the Nvidia Isaac GR00T software stack already configured. This includes the GR00T 1.7 open vision-language-action model, which helps the robot understand and act on visual and verbal instructions. It also comes with whole-body control software and a reference workflow for demonstration capture and data generation.

Researchers can interact with the robot using Nvidia Isaac Sim and Isaac Lab by using provided URDF and USD assets. These are standard file formats used to describe 3D objects and their physical properties in a simulation. The modular design allows teams to use the entire workflow or pick specific components to combine with their own custom software.

Deepu Talla, vice president of robotics and edge AI at Nvidia, stated, "Agile ONE SN combines Nvidia Isaac GR00T and Jetson Thor in a single system." This integration is meant to reduce the manual work required to validate and deploy new humanoid skills.

Path to industrial use

While the Agile ONE SN is a compact research platform, it serves as a gateway to larger applications. Agile Robots noted that skills and data developed on this smaller model can be transferred to the larger Agile ONE robot, which is built for industrial use.

The company is also exploring safety integrations, including Nvidia’s Halos safety platform and the possible use of Nvidia IGX Thor. These features would help ensure the robot operates safely in human environments. The platform is already attracting interest from major institutions, including Stanford University and ETH Zurich.

The Agile ONE SN is scheduled to become available in early 2027. By providing a complete, open hardware and software platform, Agile Robots hopes to help labs turn ideas into functional robot skills faster than current methods allow.

Data and ownership

A key consideration for research institutions is the control over their intellectual property. Agile Robots has emphasized that researchers retain control of their own data while using the platform. The open and modular nature of the system is intended to allow different labs to compare and share their work easily.

By providing a standardized starting point, the company aims to accelerate the development of Physical AI—AI that interacts with the physical world. This move helps bridge the gap between theoretical research and practical, industrial-grade robotics applications.

Technical architecture

The robot's connectivity includes Wi-Fi 7 and Bluetooth, alongside standard safety features like an emergency-stop button. The integration of Nvidia's Isaac GR00T model provides a foundation for

Future outlook

The release in 2027 marks a significant step toward making humanoid robotics more accessible to non-industrial researchers. By lowering the barrier to entry for complex robot training, Agile Robots is positioning itself as a primary infrastructure provider for the next generation of humanoid robots.