NVIDIA lays out safety stack for physical AI

Physical AI is moving toward wider commercial deployment, with ABI Research projecting an installed base of 49 million level 3-5 autonomous vehicles (AVs) by 2035 and Omdia estimating that roughly 60 million industrial robots will be deployed between 2026 and 2035. NVIDIA says safety must extend beyond a one-time predeployment check as machines begin operating on roads, in factories, in warehouses and in other spaces shared with people.

The company frames physical AI safety as a layered discipline covering hardware, software, AI behavior, operating environments and the full deployment lifecycle. It points to four shifts shaping safety models: dynamic environments that require context-aware systems, AI-specific assurance, ongoing updates and changing conditions after deployment, and validation at scale through simulation, synthetic data and scenario reconstruction.

NVIDIA Halos spans autonomous vehicle and robotics platforms, including DRIVE AGX Thor, Hyperion, Halos OS, Alpamayo, IGX Thor, Holoscan Sensor Bridge, Isaac Lab, Omniverse libraries and an outside-in safety blueprint. Partners and users named include Geely, Isuzu, Nissan, Einride, Uber, Grab, Lyft, Bosch, Gatik, Lucid, Wayve, Advantech, NexCOBOT, KION Group and Agility.

Independent assessment is also part of the strategy. TÜV SÜD certified NVIDIA’s Automotive Product Lifecycle software process and DriveOS 6.0 to ISO 26262 ASIL D, while ANAB accredited the NVIDIA Halos AI Systems Inspection Lab as an ISO/IEC 17020 inspection body.