NVIDIA Upgrades Isaac General Robotics Platform: Introduces New Humanoid Robot Reference Unit, Simulation Pipeline, Etc

　　NVIDIA announced today a major expansion of its open-source robotics development platform NVIDIA Isaac™, adding a new humanoid robot reference prototype, accompanying dataset and runtime, as well as upgrading the robotics learning framework.

　　With this technology, developers can obtain an integrated solution from data collection to deployment, enabling faster transition of general-purpose robots into real-world scenarios.

　　“General-purpose robots are rapidly advancing, and the 'ChatGPT moment' in the robotics field is already within reach,

　　Agile ONE SN —— A cutting-edge humanoid robot reference platform, aimed at research

　　Agile Robots (Agile Robots) released Agile ONE SN, which is a full-stack humanoid robot reference prototype, pre-installed with NVIDIA Isaac GR00T, equipped with the NVIDIA Jetson Thor™ edge AI platform.

　　Robot developers often face a major pain point: simulation, training, and deployment tools are fragmented, making it difficult to transfer robot skills from a development environment to a physical robot. Agile ONE SN integrates all tools and workflows into a single platform, reducing the originally weeks-long deployment preparation time to just a few days.

　　Key features of Agile ONE SN:

　　- Agile ONE S humanoid body: Height 1.4 meters, weight 40 kilograms, 31 degrees of freedom, single arm maximum payload 3 kilograms

　　- A pair of WUJI Hand 2 five-finger dexterous hands: each hand with 20 degrees of freedom, 1kHz force control, tactile perception, enabling precise manipulation

　　- On-board NVIDIA Jetson Thor computing power: FP4 computing power 2,070 TFLOPS, unified memory 128GB, used for onboard inference and real-time control

　　- Multimodal perception: Head and wrist cameras, LiDAR, inertial measurement units, perceiving and interacting with the environment

　　- NVIDIA Isaac GR00T platform: Includes GR00T base large model, complete workflow from data collection to deployment, full-body coordinated motion control

　　- Safety capabilities: Plans to integrate robot-end safety protocols, connecting to NVIDIA's full-stack safety system Halos for robots

　　“The breakthrough for humanoid robots lies in scalable practicality: a single machine that can adaptively learn multiple skills and work in various environments designed for humans,” said Chen Zhaopeng, founder and CEO of Sizheng Robotics. “Today, Sizheng is launching the Agile ONE SN, a full-stack open-source humanoid reference platform based on NVIDIA Isaac GR00T and equipped with Jetson Thor, providing developers with a ready-to-use system to develop, train, and validate the capabilities needed for the practical deployment of humanoid robots.”

　　Agile ONE SN is expected to launch in early 2027, targeting cutting-edge humanoid robot development, with top institutions such as Stanford University and ETH Zurich expressing collaboration intentions.

　　NVIDIA's new technology accelerates robot learning: from data to deployment

　　NVIDIA's new toolchain helps developers generate training data, scale robot learning, build simulation environments, and deploy robot applications, shortening the R&D cycle.

　　- SimReady Central Resource Library: provides over 2000 robots and industrial assets directly usable for simulation; Human 1M dataset includes rich human motion, synchronized multi-view video, and 3D hand tracking data. Both are now available on Hugging Face; the complete Human 1M dataset is only available to Cosmos Alliance partners.

　　- NVIDIA Isaac Lab 3.0 has officially launched: an open-source intelligent agent learning framework, supporting workflows from local workstations to data centers, multi-GPU parallel processing, and training and iterating robot strategies using Newton and NVIDIA PhysX™ physics engines.

　　- NVIDIA Isaac OS: an open-source runtime framework built on NVIDIA Holoscan, connecting AI, sensors, control units, and robot hardware to achieve deterministic real-time scheduling from simulation to physical robots.

　　- NVIDIA has released three open-source reference workflows for DGX Station™: iterating the GR00T 1.7 model using reinforcement learning, training dexterous manipulation skills in Isaac Lab, fine-tuning NVIDIA Cosmos™ world models, and generating synthetic data with ChatGPT Astra.

　　- NVIDIA Halos: a full-stack security platform based on NVIDIA IGX Thor™, helping autonomous vehicles and robots transition from prototypes to safe, scalable deployment.

　　Assets in the SimReady resource library on Hugging Face come from: Comau, FANUC, ForgeCAD, Franka Robotics, LG Electronics, Palatial, Robotiq, Standard Bots, Universal Robots, and Vertiv.

　　NVIDIA's open-source robot learning reference workflows running on DGX Station have received support from vendors such as 1X, Sinlin Robotics, Humanoid, Maven Robotics, Mentee, MicroAGI, Rhoda, Sereact, and Skild AI.

　　Multiple robot partners are integrating Halos into their safety architectures: Sinlin Robotics, Agility, Dexory, FORT Robotics, KION, for warehouse robot safety solutions; SICK is evaluating Halos to build dynamic safety capabilities for industrial robots, breaking the limitations of fixed safety zones and addressing dynamic scenarios.

　　Advantech and NexCOBOT are developing industrial computing platforms based on NVIDIA IGX; Reynolds & Moore, Accenture, and Suddeutsche TÜV are accelerating the certification process for industrial physical AI safety deployment.

　　European robot manufacturers are conducting R&D based on NVIDIA's open-source technology

　　ABB Robotics has integrated the NVIDIA Omniverse™ library into RobotStudio HyperReality, completing robot system simulation and virtual commissioning before production, covering multiple industries. KUKA Group (KUKA Group) has integrated the Omniverse component into Visual Components FactoryLens, adding real-time rendering to factory simulation. Sereact has trained and fine-tuned the Cortex platform for goods sorting using Isaac GR00T + Cosmos Transfer, deployed on Universal Robots industrial manipulators.

　　Humanoid has completed robot behavior development and validation before industrial pilot testing using Jetson Thor, Holoscan Sensor Bridge, Halos, and Isaac OS; Hexagon Robotics has developed AEON perception capabilities, including teleoperation video, localization, mapping, and object pose detection; Bear Robotics is accelerating KR1 perception and model execution, fine-tuning Isaac GR00T with Kino-2 for warehouse, return, and industrial scenarios.

　　Toyota Material Handling Europe, SoftServe use Omniverse + NVIDIA Isaac Sim™ to develop autonomous material handling solutions, with deployment cycles shortened by up to 50%. ADI>Analog Devices (ADI) uses Jetson AGX Orin™, cuVSLAM library to develop autonomous mobile robots.

　　Hugging Face and Pollen Robotics build Microduck, a small open-source robot, based on Isaac Lab, Isaac Sim, and Newton, for developers and students to train, modify, and share new embodied behaviors. Franka Robotics integrate NVIDIA NIM™ microservices + Isaac Sim to convert natural language instructions into actions for Franka FR3 Duo robotic arms. Bellboy relies on Jetson and GR00T to drive robot Billie, as its first physical AI robot partner, already deployed in hotels under Marriott, Hilton, InterContinental, and OYO.

　　Flexion uses Isaac Lab, Newton to scale training for the unified robot intelligence model Reflect; Reflect combines reasoning and full-body movement, performing complex operations across different robot models. Schaeffler (Schaeffler) exclusively provides actuator technology to Newton, combined with Isaac Lab, Isaac Sim, to accelerate humanoid robot development.

　　NEURA Robotics uses Isaac Sim, Isaac Lab to generate synthetic training data, training robot skills for multimodal foundational model Neuron AI in simulation environments.

Massive information, precise analysis, all within the Sina Weibo APP

Editor: Liu Mingliang