NVIDIA announced Halos for Robotics, the first full-stack safety system that brings proven autonomous vehicle technology to industrial robots and humanoids working alongside humans in factories and warehouses. The platform combines industrial-grade AI compute, safety-focused software, and sensor integration into a single architecture designed to help robots operate safely in dynamic, unstructured environments.

What Makes This Different From Traditional Robot Safety?

For decades, robots worked in cages, separated from human workers. But the next generation of robots, especially humanoids, need to operate in the same spaces as people. Traditional safety systems built for structured, predictable environments cannot handle the complexity of real-world factories and warehouses where conditions constantly change.

NVIDIA's approach is to leverage its deep expertise in autonomous vehicle safety. The company has accumulated over 18,600 engineering years of safety research, assessed more than 21 billion safety transistors, and produced over 7 million lines of safety-assessed code. Rather than starting from scratch for robotics, NVIDIA extended this proven foundation to physical AI systems.

"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," said Deepu Talla, vice president of robotics and edge AI at NVIDIA.

Deepu Talla, Vice President of Robotics and Edge AI at NVIDIA

How Does the Halos for Robotics System Work?

The system is organized into three interconnected layers that work together to keep robots safe:

• Hardware Platform: NVIDIA IGX Thor provides industrial-grade AI compute with built-in safety features, including a dedicated Safety Island with its own processing power, memory, and power supply isolated from the main computer. The platform can deliver up to 2,070 FP4 TFLOPs (a measure of AI processing speed) and includes over 22,000 safety mechanisms for diagnostic coverage.
• Software Stack: Halos OS runs on IGX Thor and includes Halos Core, the base safety operating system, plus specialized blueprints like the Outside-In Safety Blueprint. This software layer extends robot perception using external cameras and AI agents to dynamically control robot behavior in real-world conditions.
• Sensor Integration: Holoscan Sensor Bridge connects cameras, lidar, and other sensors to the safety compute domain over Ethernet, enabling real-time, multimodal sensor data to flow securely to the robot's decision-making system.

The IGX Thor hardware includes specific safety features that distinguish it from general-purpose computing platforms. It meets IEC 61508 SIL 3 standards, a rigorous international safety certification. The platform also includes diversity and redundancy, meaning multiple processing engines can be paired together so that if one fails, another can take over.

Who Is Using This Technology First?

Agility Robotics, a company known for its humanoid robot Digit, is the first to incorporate elements of Halos for Robotics into its proprietary safety system. Digit is designed for industrial work in logistics, manufacturing, and warehouse operations, with customers including Amazon, GXO, Schaeffler, and Toyota Motor Manufacturing Canada.

For Digit, NVIDIA IGX Thor provides the industrial-grade AI compute with built-in safety, while Halos Core supports the software layer for safety-related functions. Agility is also participating in the NVIDIA Halos AI Systems Inspection Lab, a testing and certification program that helps ensure robots meet rigorous international safety standards before deployment.

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

Peggy Johnson, CEO of Agility

What Does Certification Mean for Robot Deployment?

One of the most significant aspects of Halos for Robotics is the NVIDIA Halos AI Systems Inspection Lab, an accredited inspection body recognized by ANSI (American National Standards Institute). This lab helps companies prepare their robot systems for third-party certification by leading certification bodies including TÜV Rheinland, UL Solutions, TÜV SÜD, exida, SGS, and CertX.

The lab evaluates robots against international safety standards like IEC 61508 (functional safety), ISO 13849 (safety of machinery), and ISO/IEC TR 5469 (AI safety). Having a standardized, accredited pathway to certification significantly reduces the time and cost for companies to bring safe robots to market. The lab currently includes more than 40 companies across manufacturers, certification bodies, and safety vendors.

Steps to Prepare for Halos-Based Robot Deployment

For companies considering adopting Halos for Robotics, the path forward involves several key phases:

• Hardware Selection: Choose NVIDIA IGX Thor or compatible embedded systems from partners like Advantech and NexCobot that are designed with Halos integration in mind.
• Software Integration: Implement Halos OS and Halos Core, with support from real-time operating system partners like FreeRTOS and QNX that handle the embedded software layer.
• Safety Validation: Work with the NVIDIA Halos AI Systems Inspection Lab to test your robot's safety-related software, AI components, and cybersecurity protections against recognized standards.
• Third-Party Certification: Engage with certification bodies like TÜV Rheinland or UL Solutions to obtain formal certification before deploying robots in industrial environments.

NVIDIA has made Halos Core available in early access for registered developers in Linux and Linux plus QNX OS for Safety 8.0 configurations. The open-source Halos Outside-In Safety Blueprint is also available in early access on GitHub.

Why Does This Matter for the Future of Manufacturing?

The rise of physical AI represents a fundamental shift in how factories and warehouses operate. Robots are no longer confined to isolated cells; they work alongside human workers, sharing the same spaces and responding to dynamic conditions. This requires safety systems that are not just reactive but predictive, using AI to anticipate hazards and adjust behavior in real time.

By extending proven autonomous vehicle safety research to robotics, NVIDIA is helping establish a common safety architecture across the industry. This standardization reduces fragmentation, accelerates certification, and ultimately brings safe, capable robots to market faster. The ecosystem of partners, from semiconductor makers like Infineon and NXP to industrial application developers like FORT Robotics and Lyte AI, shows that the industry is ready to move beyond ad hoc safety implementations toward a shared, standards-aligned foundation.

As humanoid robots and autonomous mobile robots (AMRs) become more prevalent in industrial settings, the safety systems that govern them will be as important as the AI that powers them. Halos for Robotics represents a significant step toward making that safety infrastructure robust, transparent, and internationally recognized.