NVIDIA Halos OS Brings AV-Grade Safety to Robotics
Caroline Bishop Jun 22, 2026 14:01
NVIDIA launches Halos OS for robotics, extending AV-grade safety to industrial robots and physical AI, built on IGX Thor and Halos Core.
NVIDIA has unveiled Halos OS for Robotics, a full-stack safety system designed to accelerate the deployment of industrial robots and autonomous mobile robots (AMRs). Based on its proven success in autonomous vehicle (AV) safety, Halos OS extends NVIDIA's expertise into the industrial robotics sector, marking a significant step forward in physical AI safety.
The system leverages NVIDIA’s IGX Thor hardware and Halos Core software, combining real-time AI perception with built-in functional safety. IGX Thor, at the heart of this platform, delivers up to 2,070 FP4 TFLOPs of AI performance, 14 ARM CPU cores, and 128 GB of memory, offering enough compute power to handle demanding robotics workloads while simultaneously running safety monitoring systems.
Halos Core sits atop this hardware, providing a safety-certified OS that enables safe robot operation in unstructured environments like warehouses, factories, and hospitals. Two configurations are available: Linux-only or Linux with QNX, offering flexible options for developers targeting use cases that require high safety integrity.
Robotics Safety Borrowing from AV Expertise
NVIDIA’s move into robotics safety builds on over 18,000 engineering years invested in AV safety, including more than 7 million lines of safety-assessed code and over 330 published research papers. By extending these efforts into robotics, NVIDIA avoids reinventing the wheel, instead adapting proven safety processes and standards such as ISO 26262 and IEC 61508.
Agility Robotics, known for its humanoid robot Digit, has already adopted NVIDIA’s Halos OS. The company is using the platform to enhance its human detection systems, a critical component for deploying robots alongside workers safely. This partnership exemplifies the industry’s shift toward standardized safety frameworks rather than ad hoc solutions.
Key Features of Halos OS
Halos OS includes a set of safety blueprints like the "Outside-In Safety Blueprint," which integrates external infrastructure cameras and AI agents to extend a robot's perception beyond onboard sensors. This setup allows robots to dynamically adapt to complex environments, ensuring safe operation even in high-risk scenarios.
One notable use case is automated trailer loading (ATL) in warehouses, where outside-in safety enables forklifts to operate efficiently without frequent stops caused by onboard sensor limitations. NVIDIA’s Safety Decision Maker (SDM), a core component of Halos OS, ensures that robots can safely mute or reactivate onboard safety systems based on real-time data from external cameras and perception AI.
Certification Streamlined with NVIDIA Halos AI Systems Inspection Lab
To simplify compliance with global safety standards, NVIDIA operates the Halos AI Systems Inspection Lab, which pre-assesses Halos stack elements like IGX Thor and Halos Core. Companies like Agility Robotics can use this certification pathway to expedite regulatory approvals, saving significant time and costs during product development. The lab, accredited by the ANSI National Accreditation Board, supports standards such as IEC 61508 and ISO 13849.
Market Context and Implications
The launch of Halos OS for Robotics comes as NVIDIA continues to dominate the AI hardware and software markets, with its stock price recently trading at $210.69 as of June 22, 2026. The company’s push into robotics safety aligns with broader trends in industrial automation and physical AI, sectors expected to grow exponentially as technology matures.
For investors, NVIDIA’s ability to extend its AV safety expertise into robotics positions it well to capture a significant share of the $5 trillion physical AI market, which spans everything from warehouse automation to healthcare robotics. By standardizing safety, NVIDIA is not just enabling robotics innovation but also creating barriers for competitors in this space.
What’s Next?
Early access to Halos OS is now available for developers, who can start building safety applications using tools like the Halos Outside-In Safety Blueprint. NVIDIA has also made reference implementations open-source, allowing teams to quickly adapt the platform to their needs. With its proven track record in functional safety and a growing ecosystem of partners, NVIDIA is poised to set the benchmark for safe robotics deployments in the years ahead.
Image source: Shutterstock.jpg)
