AI controller translates VR, video and language commands into humanoid robot actions
Sep 11, 2026 ##### Could AI kill humanity? Experts debate risks as advanced models gain hacking capabilities Sep 11, 2026 ##### Cool, power, move: A printable textile that does it all Sep 11, 2026 ### Related Stories ##### 'Touch dreaming' helps humanoid ro...

Key takeaways
- In September 2026 a wave of high‑profile announcements underscored the rapid expansion of humanoid robotics.
- NVIDIA unveiled “SONIC,” an AI controller that converts VR, video and natural‑language commands into robot actions, promising to broaden the motion repertoire of humanoids, while related work on “touch‑dreaming” techniques lifted task‑success rates by over 90 percent earlier in the year.
- Goldman Sachs dramatically raised its market outlook, projecting shipments of roughly 6.5 million humanoid robots by 2035—a five‑fold jump from its previous estimate—and valuing the sector at about $138 billion.
- In China, the People’s Liberation Army allocated about $300,000 in June to collect and label sensor data for fielded robots, and Reuters reported that the PLA is urging researchers to accelerate the transition of commercial humanoids into combat roles.
- That push coincided with Xpeng’s release of a video on September 11 showing its IRON humanoid walking off an assembly line unaided, and a Chinese startup, JoyIn, claiming its “Aether” model, built on a perceptive‑control approach, can achieve a 90 percent first‑attempt success rate—a claim it says mirrors OpenAI’s technology.
Sep 11, 2026
Could AI kill humanity? Experts debate risks as advanced models gain hacking capabilities
Sep 11, 2026
Cool, power, move: A printable textile that does it all
Sep 11, 2026
'Touch dreaming' helps humanoid robots handle five tricky tasks with 90.9% higher success
May 12, 2026
Using hierarchical generative models to enhance the motor control of autonomous robots
Dec 10, 2023
Learning dance moves could help humanoid robots work better with humans
Jul 11, 2024
Robots learn to perceive objects using acoustic vibrations
Oct 23, 2024
A scalable reinforcement learning–based framework to facilitate the teleoperation of humanoid robots
Apr 6, 2024 Humanoid robots, robotic systems with limbs and body structures that resemble those of humans, could tackle various manual tasks in homes, workspaces and other settings. Yet teaching these robots to reliably perform different humanlike movements is typically challenging and time-consuming.
Researchers at NVIDIA recently developed a controller, an artificial intelligence (AI) model that translates movement targets into commands that can be executed by a robot, that could help expand the motion repertoire of humanoid robots. The new controller, dubbed SONIC, was described in a paper published in Science Robotics. Apr 6, 2024
Human-aware robots adapt to partners, reducing back strain during team lifting
Aug 5, 2026
Big AI wants to slow down AI research: Is it a safety pause or a strategic retreat?
2 hours ago
Is it possible to slow down the development of AI?
6 hours ago
Anthropic CEO Dario Amodei says AI industry needs to give safety measures time to catch up
Sep 13, 2026
Underwater solar cells can operate 10 meters below the ocean's surface
Sep 11, 2026
Weighing 600 million buildings to help construct a more efficient future
Sep 11, 2026
Laser annealing integrates optical isolators directly onto silicon photonics chips without overheating
Sep 11, 2026
Get Instant Summarized Text (Gist)
New robots, funding, and signals. Every Thursday.
Unsubscribe anytime.
Mentioned in this article