Skip to content

Artificial Intelligence

NVIDIA Partners With Comau and LG Innotek on Isaac Sim Motion Controllers and Holoscan Humanoid Vision Bridges

GTC Berlin physical AI announcements embed Jetson Thor compute and low-latency sensor pipelines into industrial automation and bimanual robots.

By Lazy Tensor
Published · 2 min read

Dual articulated industrial robotic arms with multi-camera vision heads inside a smart factory lab beside a 3D digital twin simulation display
Dual articulated industrial robotic arms equipped with stereo vision camera arrays operate beside an Isaac Sim 3D digital-twin workstation.

Hardware-accelerated physical AI took center stage in late October 2026 as NVIDIA unveiled production partnerships with industrial automation pioneer Comau and optical engineering giant LG Innotek, accompanied by sweeping semiconductor packaging agreements affecting Silicon Valley, San Jose, and global AI data center supply chains. Announced during the GTC Berlin physical AI summit on Thursday, October 22, 2026, the official developer program releases push synthetic simulation and ultra-low-latency sensor fusion directly into factory-floor motion controllers and humanoid robot heads.

Under the Comau collaboration, the Italian industrial robotics manufacturer is embedding NVIDIA Isaac Sim Omniverse digital-twin pipelines and Jetson edge AI compute modules directly into its next-generation industrial robot controllers. Instead of relying on brittle hand-coded programmable logic controller (PLC) waypoints, automotive and electronics assembly engineers in San Francisco Bay Area and European plants can train collision-aware grasping and multi-arm weld paths in physically accurate simulation before deploying compiled policies directly to shop-floor actuators.

Addressing the sensory bottleneck in general-purpose humanoid and mobile robots, LG Innotek partnered with NVIDIA to integrate Holoscan Sensor Bridge (HSB) Ethernet-to-GPU streaming into its upcoming multi-modal RGB-D, thermal, and Time-of-Flight (ToF) camera modules. By bypassing traditional CPU interrupt overhead and DMA copies, Holoscan Sensor Bridge delivers sub-millisecond frame ingestion into Jetson Thor vision-language-action (VLA) transformer models.

To ensure silicon supply keeps pace with both cloud training clusters and edge robotics deployments, NVIDIA also confirmed multi-year capacity agreements scaling advanced ABF substrate output 2.5x with IBIDEN, Unimicron, and Samsung Electro-Mechanics, alongside a 15x co-packaged optics ramp with SPIL and a $1.1 billion U.S. wafer-testing reservation with KYEC. Robotics developers in San Jose and San Francisco can access updated Isaac ROS and Holoscan SDK containers through the official NVIDIA developer program at developer.nvidia.com/isaac.

Sources

Comments (0)

No comments yet. Be the first to share a thought on this story.

The free newsletter

Silicon Valley, in your inbox.

The local stories worth knowing, delivered once a week.

From Silicon Valley, San Francisco, San Jose, Palo Alto, Mountain View, and Berkeley.

Get the newsletter

Free to read. Unsubscribe anytime.

More local stories