News

NVIDIA Pitches Safety Stack for Robots Moving into Factories and Warehouses

• By John K. Waters
• 06/22/2026

NVIDIA is expanding its robotics strategy with a safety system aimed at companies building AI-enabled machines for factories, warehouses, and logistics sites, where robots are expected to operate near people, equipment, and other automated systems.

The company said Monday that NVIDIA Halos for Robotics is a full-stack safety system for “physical AI,” a term NVIDIA uses for machines that can sense, decide, and act in the real world. NVIDIA described the system as a common safety architecture spanning AI compute, system software, sensor data, safety applications, and inspection.

The announcement is less about a single chip or robot than an effort to make NVIDIA a central infrastructure provider for robotics safety. As companies move from controlled demonstrations to real industrial deployments, robot developers face pressure to show that autonomous systems can be monitored, validated, and certified before they operate at scale.

“Physical AI is transforming how factories, warehouses, and logistics operations work,” Deepu Talla, vice president of robotics and edge AI at NVIDIA, said in the company’s announcement. He said robotics teams need “a unified safety architecture” to scale autonomous systems in those settings.

NVIDIA said Halos for Robotics includes NVIDIA IGX Thor and Holoscan Sensor Bridge for industrial AI compute and sensor connectivity. The software layer includes Halos OS, Halos Core, and an Outside-In Safety Blueprint, which NVIDIA said uses external cameras and AI agents to help control robot behavior in industrial spaces.

The company also said its NVIDIA Halos AI Systems Inspection Lab is intended to help partners prepare for third-party certification. NVIDIA described the lab as an ANSI National Accreditation Board-accredited program for functional and AI safety for physical AI. The board is part of the American National Standards Institute, commonly known as ANSI.

Agility, maker of the Digit humanoid robot, is the first company NVIDIA named as using the new robotics safety system. NVIDIA said Agility will integrate NVIDIA IGX Thor and Halos Core into its safety system for Digit, which is designed for logistics, manufacturing, and warehouse operations.

Agility will also work with NVIDIA’s inspection lab to evaluate Digit’s safety-related software, AI components, and cybersecurity protections against standards including IEC 61508, ISO 13849, and ISO/IEC TR 5469 before final third-party certification, NVIDIA said.

“For humanoids to deliver value at scale, safety has to be built into the robot,” Agility Chief Executive Peggy Johnson said in NVIDIA’s announcement.

NVIDIA also framed Halos for Robotics as an ecosystem effort. The company said certification bodies, including TÜV Rheinland, TÜV SÜD, UL Solutions, exida, SGS, and CertX, recognize the NVIDIA Halos AI Systems Inspection Lab as part of their certification process.

The company said Halos Core for NVIDIA IGX is available in early access for registered developers in Linux and Linux plus QNX OS for Safety 8.0 configurations. NVIDIA also said the open-source Halos Outside-In Safety Blueprint is available in early access on GitHub.

The timing reflects a shift in robotics from isolated automation toward systems that must operate in less predictable environments. Industrial robots have long worked behind cages or within tightly controlled workflows. Humanoids, mobile robots, and other physical AI systems are expected to operate in environments where people, vehicles, shelves, conveyors, and machines may all be moving simultaneously.

That creates a different kind of safety problem. It is not enough for a robot to be powerful or accurate. It must also be able to perceive its environment, respond to unexpected conditions, stop safely, protect workers, resist cybersecurity failures, and pass external review.

NVIDIA is positioning Halos for Robotics to address that problem across the stack. The company’s pitch is that robot makers should not have to build safety architecture separately for compute, sensors, software, and certification.

The practical question is whether robotics companies, customers, and certification bodies will converge around NVIDIA’s approach. The announcement shows NVIDIA trying to turn robot safety into a platform layer, much as it has done with accelerated computing in AI data centers and autonomous vehicle development.

For robotics developers, that could simplify the path from prototype to deployment. For customers, it could provide a clearer safety framework for evaluating robots before they are placed near workers. But the broader challenge remains: AI-enabled machines will have to prove themselves not only in benchmarks and demos, but in the varied, messy, and regulated environments where industrial work actually happens.