content image

チャットボットからヒューマノイドへ:NVIDIA GTC 2026、フィジカルAIの時代を切り開く

From Chatbots to Humanoids: NVIDIA GTC 2026 Ushers In the Era of Physical AI

NVIDIA GTC 2026で掲げられた「Physical AI」構想。AIがチャットボットを超え、現実世界を理解し行動する時代へ——その全貌と最新技術を紹介する。
分からないところをタップすると
↓日本語訳が表示されます↓

At NVIDIA GTC 2026, held in San Jose from March 16 to 19, more than 30,000 people from over 190 countries gathered to see AI move beyond chatbots and image tools. In Jensen Huang’s keynote, one message stood out: the next wave is “physical AI,” meaning systems that can understand the real world and act inside it. That idea became vivid when Olaf from Disney’s Frozen appeared onstage as a physically simulated character, suggesting a future where intelligence does not stay trapped on a screen. (nvidianews.nvidia.com)

What is speeding up this shift is not just better robots, but a complete training pipeline. NVIDIA has released open models such as Cosmos Reason 2 for physical-world reasoning and Isaac GR00T N1.6 for humanoid robots, along with Isaac Lab-Arena for large-scale testing in simulation and OSMO for managing robotics workflows across cloud and local computers. NVIDIA also said its GR00T and Isaac tools are being integrated into Hugging Face’s LeRobot ecosystem, while Jetson Thor is becoming a key onboard computer for humanoids from companies including Boston Dynamics, Humanoid and NEURA Robotics. (investor.nvidia.com)

At GTC 2026, the practical side was just as important as the vision. Texas Instruments showed a system that combines radar with Jetson Thor and Holoscan to give humanoids low-latency 3D perception. This matters because radar can detect difficult obstacles, including reflective or transparent surfaces such as glass doors, more reliably than cameras alone. ABB, meanwhile, announced RobotStudio HyperReality, built with NVIDIA Omniverse libraries. ABB says it can reduce deployment costs by up to 40%, speed time to market by up to 50%, and achieve 99% correlation between simulation and real-world behavior. (ti.com)

The most interesting point is that AI is no longer only something we talk to. At GTC 2026, it looked more like a new industrial system: world models that teach robots, simulations that replace expensive trial and error, sensors that improve safety, and edge computers that help machines act in real spaces. Humanoids are still early, but the direction is becoming clearer. AI is learning not only to answer questions, but also to move through the world. (blogs.nvidia.com)

by EigoBoxAI
作成:2026/03/21 03:01
レベル:中上級 (語彙目安:4000〜6000語)

まだ読んでいないコンテンツ

content image
by EigoBoxAI
作成:2026/03/21 03:04
レベル:超上級 (語彙目安:8000語以上)
content image
by EigoBoxAI
作成:2026/03/20 21:03
レベル:超入門 (語彙目安:〜300語)
content image
by EigoBoxAI
作成:2026/03/20 21:01
レベル:初級 (語彙目安:300〜1000語)
content image
by EigoBoxAI
作成:2026/03/20 21:00
レベル:中級 (語彙目安:2000〜2500語)
content image
by EigoBoxAI
作成:2026/03/20 15:03
レベル:初中級 (語彙目安:1000〜2000語)
content image
by EigoBoxAI
作成:2026/03/20 15:02
レベル:超上級 (語彙目安:8000語以上)
content image
by EigoBoxAI
作成:2026/03/20 15:01
レベル:上級 (語彙目安:6000〜8000語)
content image
by EigoBoxAI
作成:2026/03/20 09:05
レベル:中上級 (語彙目安:4000〜6000語)
content image
by EigoBoxAI
作成:2026/03/20 09:03
レベル:超入門 (語彙目安:〜300語)