NVIDIA has introduced what it calls the industry’s first full-stack safety system purpose-built for robots and physical AI, a platform called Halos for Robotics that is designed to keep autonomous machines operating safely in factories, warehouses and logistics facilities where humans are also present.
What Halos for Robotics does
At Automate 2026 in Chicago, NVIDIA announced Halos for Robotics as the industry’s first full-stack, comprehensive safety system for robotics and physical AI that unifies AI compute and safety. The announcement came on June 22. Automate 2026 ran June 22, 25 at McCormick Place in Chicago, North America’s largest robotics and automation event.
The platform integrates NVIDIA IGX Thor compute, the Holoscan Sensor Bridge, Halos OS and the Halos AI Systems Inspection Lab. The important shift is standardization: safety for physical AI is moving from ad hoc application logic toward a stack that combines compute, sensors, system software, validation and inspection.
The system draws on more than 18,600 engineering years of safety expertise from NVIDIA’s autonomous driving domain and extends it to the humanoid robots now entering industrial settings. NVIDIA describes Halos for Robotics as the industry’s only full-stack, open robotics safety system, giving machines that sense, decide and act in the real world a single common safety architecture.
Agility’s Digit becomes the first adopter
Agility, a leading humanoid robotics and physical AI company, is the first to use NVIDIA Halos for Robotics to build safety into its humanoids working in factories, warehouses and logistics operations for customers including Amazon, GXO, Schaeffler and Toyota Motor Manufacturing Canada.
Agility plans to integrate NVIDIA IGX Thor and Halos Core into its existing human detection and safety systems for the Digit humanoid robot, which is designed for industrial work in logistics, manufacturing and warehouse operations.
As humanoid deployments scale, that human-detection capability is a prerequisite for mixed-workforce environments, where robots must reliably distinguish a worker in their path from a static obstacle.
Agility will also participate in the NVIDIA inspection program to evaluate software, computing systems and cybersecurity protections against standards including IEC 61508, ISO 13849 and ISO/IEC TR 5469 before final third-party certification.
Agility CEO Peggy Johnson said: “For humanoids to deliver value at scale, safety has to be built into the robot and validated across the entire system.” She added that partnering with NVIDIA to implement and optimize Halos for Robotics extends the company’s commitment to responsible automation, which she described as a non-negotiable requirement for scaling humanoid deployments.
Developer access and broader ecosystem
NVIDIA Halos Core for NVIDIA IGX is available in early access for registered developers in Linux and Linux plus QNX configurations. The open-source NVIDIA Halos Outside-In Safety Blueprint, part of the Halos Applications layer of Halos OS, is now available in early access on GitHub.
The safety architecture matters in the context of a fast-expanding market. According to new data released by the Association for Advancing Automation, North American companies ordered 8,940 robots valued at $622 million in the second quarter of 2026. Second-quarter results brought first-half totals to 17,995 units valued at $1.166 billion.
Non-automotive customers accounted for 56 percent of second-quarter robot orders, a sign that the industries most likely to deploy humanoids alongside human workers are also driving current volume. For context on how quickly AI capabilities are now being embedded into hardware across industries, GCN previously reported on robot vacuums losing AI features when subscriptions lapse, illustrating how tightly software and safety are now intertwined with physical machines.
Why safety certification matters now
Physical AI safety is becoming a platform problem rather than a feature teams can bolt on after a robot works in a demo. Regulators and enterprise buyers are increasingly demanding documented, third-party-verified safety records before approving humanoids for production floors.
The Halos AI Systems Inspection Lab provides pre-assessed safety and AI certification pathways for partners, reducing time and cost for third-party system certification to standards like IEC 61508 and ISO 13849. Partners across the ecosystem include Agility, Boston Dynamics and Hesai, among more than 43 companies that have joined the Halos ecosystem.
For the robotics industry, Halos represents an attempt to do for humanoid safety what standardized crash-test ratings did for automotive, establishing a shared benchmark that operators, insurers and regulators can rely on as humanoid deployments move from controlled pilots to the open factory floor. Those following NVIDIA’s broader industrial AI buildout can track related developments in the company’s PAIR software push for distributed AI agents, which signals how deeply the company is embedding itself across every layer of the physical AI stack.
