What happened
At GTC Berlin, NVIDIA expanded its Isaac robotics platform with humanoid reference platforms, datasets and a new execution framework, saying setup time drops from weeks to days.
Why it matters
If that holds, robot makers could move from simulation to real machines far faster, lowering the time and cost of developing general-purpose robots.
WHO IT HITSRobot makers and robotics software developers stand to gain from a more integrated toolchain, while enterprises planning humanoid pilots may see faster iteration if the setup-time claim holds.
Summaries like this, in your inbox every morning.
NVIDIA used its GTC Berlin stage to broaden Isaac from a development toolkit into a fuller stack for building general-purpose robots. The additions span simulation-ready assets, training frameworks and a real-time execution layer, aiming to connect data generation, learning, simulation and deployment in one workflow. A key claim is that this integration cuts setup time from weeks to days.
The most concrete example comes from Agile Robots, which pre-configured Isaac GR00T and Jetson Thor into the Agile ONE SN, a humanoid built on the Agile ONE S chassis with 31 degrees of freedom and 3 kg payload per arm. Its WUJI Hand 2 features 20 degrees of freedom per hand with 1 kHz force control and tactile sensing. Jetson Thor provides 2,070 TFLOPS at FP4 and 128 GB of unified memory for on-robot inference. The system is planned for early 2027.
On the software side, NVIDIA released Isaac Lab 3.0 for general availability and introduced Isaac OS, an open-source runtime built on Holoscan. It also published over 2,000 simulation-ready assets and the Human 1M Dataset, with full access for Cosmos Coalition partners. Together, these pieces suggest NVIDIA is trying to standardize how robot makers train and deploy models, though the setup-time claim remains to be validated in practice.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Bear Robotics, LG Electronics' robotics subsidiary, is adding Chinese manufacturers to its South Korean outsou…
Tesla reported third-quarter capital expenditures of $4.73 billion, short of the $7.95 billion analysts expect…
Tesla's third quarter showed its vehicle business regaining some per-car profitability, even as total sales st…
BlackRock added a robotics ETF, expanding its roster of AI-themed products, Reuters reported
Mitsubishi Electric said on October 21 it developed a system in which humanoid robots and CNC machine tools co…
Mitsubishi Electric says it developed an industry-first system where a humanoid robot and a CNC (Computer Nume…
