Agile Robots launches Nvidia-powered robotics platform ahead of early 2027 availability
Key Facts
Agile Robots has launched the compact "Agile ONE SN" humanoid research platform, equipped with Nvidia technologies, which it says cuts setup time from weeks to days by bringing robotics development tools together. The platform comes preconfigured with Nvidia's "Isaac GR00T" platform and runs on "Jetson Thor" edge computing, according to a statement from the U.S. company. This configuration brings simulation, training and deployment tools together on the new platform to support researchers and developers working on robotics development. Nvidia described the effect of unifying these tools as "reducing setup time from weeks to days". The preconfiguration thus includes the robotics development platform and the edge computing on which it runs, as part of the configuration announced by Nvidia. The platform's research use is linked to a pathway announced by Agile Robots for subsequently transferring skills and data developed on it to its industrial applications.
Agile Robots positions the compact research platform alongside "Agile ONE", its full-sized counterpart designed for industrial applications within the company's portfolio. It said in its statement that "skills and data developed on Agile ONE SN can subsequently be transferred to Agile ONE". This relationship allows the compact platform to be used to develop skills that can be transferred to the larger robot intended for industrial work, according to the company's description. Its sensing equipment includes cameras in the head and wrists, along with lidar sensors to help the robot perceive its surroundings, according to Nvidia. In a separate industrial application, Audi, Agile Robots, Dassault Systèmes and Nvidia announced a proof-of-concept project to automate parts sequencing at Audi's Ingolstadt plant using mobile humanoid robots. The project targets handling varied or semi-flexible parts and reliably picking disordered components, two persistent automation challenges identified by Audi.
Agile Robots' humanoid robot manufacturing operations are based in a 10,000-square-metre hall in Fürstenfeldbruck, according to a blog post published by the company in September. The company said production at the factory began in January, with the line entering series production by late summer. Over that period, the line thus moved from the start of production to the series manufacturing stage described by the company in its account of the factory's development. A complete robot takes 3 to 4 days to pass through the factory's stages, from the first station to the finished unit, according to the post. This period covers the unit's journey through the stations to its final form, as stated in the company's description of the manufacturing process. Agile Robots emphasises its in-house production strategy, saying: "We design and manufacture our robots ourselves."
Nvidia's ties with Agile Robots extend to planned investments announced by Jensen Huang, alongside planned investments in Avelios Medical, Mirelo and Reflex Aerospace, according to Nvidia's blog. The announcement adds to the presence of the U.S. company's technologies in the new research platform's configuration, from the robotics development platform to edge computing. Agile Robots also has an industrial partnership with Hitachi, under which it said it plans to use the Japanese company's edge AI semiconductors in its robots. Agile Robots outlined the plan by saying: "We will use Hitachi's edge AI semiconductors in our robots." Across its broader portfolio, Chief Executive Zhaopeng Chen told Reuters in an interview before the announcement that the company had more than 20,000 robots operating in various applications. That figure refers to robots operating across the company's portfolio in its different applications.
In another research initiative, Agile Robots previously announced a strategic partnership with Google DeepMind in March to integrate "Gemini Robotics" foundation models with its hardware. The research partnership brings together AI models and the robotic hardware developed by Agile Robots, according to the company's statement. The partnership relies on real-world operating data to improve the models, according to Agile Robots' explanation of its intended learning mechanism. The company said that "data from real-world operations improves the models, and improved models expand the robots' capabilities". As it describes it, this process begins with data generated by real-world operations, then moves on to improving the models and expanding the hardware's capabilities.
For the new research platform, Nvidia said Stanford University and ETH Zurich had expressed interest. The two institutions' interest comes in the context of the research and development uses targeted by the compact "Agile ONE SN" platform. Its development roadmap includes integrating safety protocols into the robot, a step Nvidia described as planned. Integrating these protocols forms part of the upcoming development work the company announced for the platform, alongside its announced simulation, training and deployment capabilities. "Agile ONE SN" is scheduled to become available in early 2027, according to the timeline announced by Nvidia.