Halos for Robotics: NVIDIA's Full Stack Robot Safety System

NVIDIA has revealed a first of its kind full-stack safety comprehensive safety architecture to robotics and physical AI systems: Halos for Robotics.

The technology applies autonomous vehicle safety principles to humanoid robots operating in industrial settings.

NVIDIA's announcement marks the first application of Halos beyond automotive use cases. Agility has integrated NVIDIA Halos for Robotics into its humanoid systems deployed at Amazon, GXO, Schaeffler and Toyota Motor Manufacturing Canada.

Safety architecture for physical AI

Halos for Robotics combines three technical layers to address safety requirements in autonomous systems.

NVIDIA IGX Thor provides industrial-grade AI compute with built-in safety capabilities while Holoscan Sensor Bridge handles sensor connectivity.

The Halos OS software stack manages safety functions at the operating system level. NVIDIA Halos AI Systems Inspection Lab offers third party certification services.

Deepu Talla, Vice President of Robotics and Edge AI at NVIDIA, says: "Physical AI is transforming how factories, warehouses and logistics operations work and robotics teams need a unified safety architecture to scale autonomous systems into these environments.

"With NVIDIA Halos for Robotics, developers and system builders can harness NVIDIA’s proven autonomous vehicle safety foundation to develop safer robots faster and bring them into industrial operations alongside workers with greater confidence."

Technical implementation in humanoid systems

Agility has deployed Digit humanoid robots using components of the NVIDIA safety system. According to Agility, Digit operates as the first humanoid robot in production deployment with Arc – a multi-cloud management platform developed by Microsoft.

Peggy Johnson, CEO of Agility, says: "For humanoids to deliver value at scale, safety has to be built into the robot and validated across the entire system.

"Partnering with NVIDIA to implement and optimise the Halos for Robotics system extends our leadership in responsible automation, which is a non-negotiable requirement for bringing humanoids safely into industrial workflows.

"This collaboration unlocks true human-robot teamwork, driving the long-term returns that will power next-generation manufacturing and logistics operations."

AI safety challenges in robotics

The expansion of autonomous systems into physical environments raises technical questions about AI model reliability.

According to the International Federation of Robotics World Robotics 2025 statistics, 542,000 industrial robots were installed in 2024 – more than double the number installed ten years prior.

Deployment challenges in autonomous vehicles have informed discussions about robotic safety systems.

In a recent interview with McKinsey, Daniela Rus, Director of MIT's Computer Science and Artificial Intelligence Laboratory, said: "AI models that control robots are typically not closed-form solutions.

"There is always a chance that they will make a mistake. How will the robot respond if the AI brain tells it to do something wrong?"

The technical challenge could mean AI systems require validation frameworks that account for non deterministic behaviour. NVIDIA's approach applies automotive safety engineering to address these uncertainties in robotic applications.