Nvidia's breakthrough in robot safety.
Nvidia Launches Halos for Robotics, Unifying AI for Safer Industrial Automation
Key Points
Nvidia Robotics/edge AI VP Deepu Talla said the platform is meant to let “developers and system builders… develop safer robots faster” and “bring them into industrial operations alongside workers with greater confidence.”
Nvidia described Halos for Robotics as unifying not just compute, sensors, and safety software, but also “inspection tools” into a single safety architecture.
Agility’s first adoption is tied to its humanoid robot Digit, which is incorporating Nvidia “IGX Thor and Halos Core” for use in factories, warehouses, and logistics for customers including Amazon, GXO, Schaeffler, and Toyota Motor Manufacturing Canada.
FORT Robotics said it will demo an “agentic safety application” built on Nvidia’s open-source “Halos Outside-In Safety Blueprint” and powered by Nvidia “IGX Thor and Holoscan Sensor Bridge,” at Automate in Chicago and alongside Nvidia in the Humanoid Robotics Pavilion.
FORT contrasted its outside-in approach with “inside-out safety systems” that rely only on onboard perception and “typically enforce conservative speed and separation limits,” aiming instead to dynamically change robot operating modes using fixed infrastructure; FORT also cited warehouse/factory use cases like trailer unloading, inventory replenishment, and assembly.
Nvidia launched Halos for Robotics on June 22 at the Automate conference in Chicago, calling it the industry's first full-stack safety platform for industrial robots. The system unifies AI computing hardware, sensors, safety software, and inspection tools into one architecture, letting robots work at full speed alongside human workers instead of constantly slowing down or stopping near them. The Robot Report reports that early adoption is already underway, with humanoid robotics firm Agility deploying the system for customers including Amazon, Toyota Motor Manufacturing Canada, GXO, and Schaeffler.
Nvidia VP of Robotics and Edge AI Deepu Talla said the platform lets "developers and system builders develop safer robots faster" and bring them "into industrial operations alongside workers with greater confidence." The company built Halos by repurposing safety technology from its DRIVE autonomous-vehicle program — a foundation that includes 18,600 engineering years of development, 21 billion safety-assessed transistors, and 7 million lines of safety-checked code, according to GuruFocus.
Agility's Digit Humanoid Is First to Adopt Halos
Agility Robotics confirmed it is the first humanoid firm to build Halos into a production robot. Its Digit robot now runs on Nvidia's IGX Thor compute hardware paired with Halos Core software. Agility CEO Peggy Johnson called safety a "non-negotiable requirement" for humanoids working in industrial settings, according to The Robot Report. The Digit robot is already operating in factories and warehouses for Amazon, GXO, Schaeffler, and Toyota Motor Manufacturing Canada.
Nvidia made Halos available in early access for registered developers and released portions as open-source software. The open-source Halos Outside-In Safety Blueprint went live on GitHub on June 17 — five days before the official launch — allowing partners to start building integrations early. GuruFocus notes that full production certification for individual customer implementations is still ongoing, making the current rollout a proving ground for 2027–2028 deployment cycles.
Fixed Cameras Beat Onboard Sensors in Busy Warehouses
A key debate in robot safety is whether to rely on sensors mounted on the robot itself — called "inside-out" safety — or on cameras fixed to the ceiling and walls, called "outside-in" safety. Traditional inside-out systems "typically enforce conservative speed and separation limits," forcing robots to crawl or stutter near workers. FORT Robotics, which is partnering with Nvidia on Halos, argues that fixed infrastructure gives a fuller picture of the workspace and keeps robots moving faster without sacrificing safety.
FORT's system uses external cameras and visual AI agents to watch the entire workspace. A central "Safety Core" can tell a robot it is clear to move even when the robot's own sensors are blocked — such as inside a dark trailer during unloading. FORT CEO Samuel Adams said the integration extends what he calls a "Trust Layer" beyond onboard sensors to the whole facility, maximizing throughput. FORT plans to demo the system at Automate in Chicago and in the Humanoid Robotics Pavilion alongside Nvidia, according to The Robot Report.
World's First Physical AI Safety Lab Backs the Platform
Nvidia also launched the Halos AI Systems Inspection Lab, which it says is the world's first lab accredited by the ANSI National Accreditation Board under ISO/IEC 17020 for physical AI inspection. The lab works with major certification bodies including TÜV Rheinland, UL Solutions, TÜV SÜD, and SGS. Developers can use it to pre-validate their safety cases covering functional safety, cybersecurity, and AI system integrity before seeking final certification, The Robot Report reported.
The platform's core hardware, the IGX Thor chip, runs a BlackBerry QNX OS for Safety 8.0 — an operating system already certified to ISO 26262 ASIL D and IEC 61508 SIL 3, the highest levels of functional safety in their respective standards. BlackBerry QNX President John Wall said "safety and determinism cannot be afterthoughts" in software-defined robotics. The partnership between Nvidia and BlackBerry QNX was first announced at Hannover Messe in April 2026, according to GuruFocus.
A $200 Billion Market Rides on Getting Safety Right
Barclays estimates the humanoid robotics market will hit $200 billion by 2035. Analysts at GuruFocus suggest Nvidia's move into the safety-certification layer positions it as the "operating system" of the future automated workforce — collecting a toll on every robot that gets certified through its platform. Amazon and Toyota's early involvement signals that large employers see certified collaborative safety as the key to moving humanoids from pilot programs to permanent roles.
The platform also signals a regulatory shift in how robot safety is judged. Industry rules are moving away from simple on-off safety switches toward AI-based reasoning that is still certifiable and auditable — a standard called SOTIF, or Safety of the Intended Functionality. The Los Angeles Times noted that Nvidia is pushing for robots to handle "split-second decisions" before they can truly work side by side with humans on factory floors.
Sentiment
Mostly PositivePublishers
20
Articles
48
Reach
68
