Robotics Safety
Physical AI demands a new standard of safety. NVIDIA Halos brings the full-stack platform proven in autonomous vehicles to the next generation of humanoids and industrial robots.
Overview
NVIDIA Halos for Robotics
NVIDIA Halos is the comprehensive safety system for physical AI that takes robots from prototype to production. Built on the same trusted foundation that powers autonomous vehicle safety, Halos extends that proven architecture to humanoids and industrial robots operating in the real world—in warehouses, factories, and alongside people.
NVIDIA delivers a full-stack safety solution spanning infrastructure, operating system, middleware, and applications, all backed by a broad ecosystem of OEMs, sensor, and software partners.
Benefits
Why Halos for Robotics?
Where AI Meets Safe Deployment
NVIDIA Halos unifies real-time AI, functional safety, and edge compute into a single safety stack. It brings together what has never been combined before to enable robots to work safely alongside humans.
AV-Proven, Robotics-Ready
NVIDIA Halos was built over years of autonomous vehicle deployment. The same safety architecture, now adapted for humanoids and industrial robots, means you're not starting from scratch. You're building on a foundation that's already been tested in the world's most demanding safety environments.
Full Stack
From NVIDIA IGX Thor™ to Halos OS to open safety application blueprints, the entire stack is engineered to work together. Partners and OEMs leverage a production-ready software stack and achieve faster time to market.
Open-Source Application for Developers
Available through GitHub, find open source blueprints and workflows to build, customize, and deploy safety agents. These include NVIDIA Halos Outside-In Safety Blueprint, a practical entry point for safety-critical robotics teams that extends perception from on-board sensors to external cameras.
A Path for Safety Certification
Halos elements undergo third-party inspection or certification by recognized notified bodies. With the Halos framework, safety certification is a structured process, not an afterthought. The Halos AI Systems Inspection Lab accelerates the time to certification of Halos customers.
Comprehensive Partner Ecosystem
Leading robotics companies, OEMs, industry safety pioneers, certification agencies, and software and sensor providers worldwide are using the system to deliver robotic safety at all levels of automation.
Technology
End-to-End Safety Stack
Halos OS integrates safety at every level of the robotics stack, including hardware, OS, middleware, and applications.
Figure 1. NVIDIA Halos for Robotics full-stack architecture spanning platform safety, Halos OS, and ecosystem/algorithmic safety layers for AI-driven robot behavior control.
Use Cases
Real-World Applications
Where Halos for robotics is working today.
-
Use Case 1
-
Use Case 2
-
Use Case 3
Inside-Out Safety: Humanoid Robot Deployment With Agility
Digit, Agility's humanoid, will be the first production robot shipping with NVIDIA Halos OS. It will operate in real warehouse environments, using NVIDIA IGX™ and Halos Core to safely navigate and work alongside people.
Outside-In Safety: Autonomous Forklift for Trailer Loading
In this safety concept, inspected by TÜV Rheinland, outside-in cameras establish a virtual fence and monitor the trailer interior and docking zones for occlusion alerts. Robots operate at full speed or high-efficiency mode when no workers are in the region of interest, and engage in safety functions the moment a person enters to prevent incidents from occurring.
Outside-In Safety: Mobile Robot Zoning
Virtual tripwires and dynamic zones manage multiple mobile robots such as AMRs in shared spaces with people in aisles and intersections. Alerts and occlusion detection are imperative when a person or robot enters an area with blind corners or high racks, maintaining safety even in blind spots.
Ecosystem
A Safety Ecosystem Built for Scale
-
Humanoid and Robot OEMs
-
Software and OS Partners
-
Hardware and Platform Partners
-
Certification and Compliance
-
Outside-In Safety Ecosystem
-
Safety MCU/HSB Partners
Resources
The Latest in NVIDIA Halos Resources
-
News
-
Sessions
-
Livestreams
Next Steps
FAQs About NVIDIA Halos
NVIDIA Halos is an end-to-end safety platform designed to take autonomous robots from prototype to production. It provides a full-stack solution spanning silicon, operating systems, middleware, and applications to support functional safety requirements.
- Learn more about the NVIDIA Halos platform and its safety architecture.
Halos OS is a safety-certified operating system foundation for robotics that runs on the NVIDIA IGX Thor platform, integrating Linux and QNX to deliver real-time control and safety extensions.
- Review the Halos OS safety foundation components.
Inside-out safety refers to on-board robot sensors managing the immediate safety envelope of the robot, whereas outside-in safety uses external infrastructure (like external cameras) to monitor the environment and establish virtual zones.
- Compare these approaches in the Robot Safety Use Cases guide.
Halos extends autonomous vehicle-proven safety architecture to humanoids and industrial robots, enabling safe navigation and operation in warehouses and factories alongside human workers.
The NVIDIA IGX is a third-party assessed, safety-compliant System-on-a-Chip that includes built-in hardware safety mechanisms, such as a Functional Safety Island (FSI), to meet stringent industrial safety standards.
- Get details on the NVIDIA IGX safety specifications.
Developers can access the SDK, along with Halos Core Linux and IGX OS, by visiting the NVIDIA robotics developer portal to begin integrating the safety stack.
- Start your development journey by accessing the NVIDIA Halos SDK.
Halos provides validated safety blueprints, certified hardware (IGX), and third-party inspection services that allow developers to transition their safety systems from initial prototyping to reliable, production-ready deployments.
- Read about the full-stack safety platform journey from prototype to production.