NVIDIA Launches First Accredited Lab for AI Safety in Autonomous Vehicles with Hesai and Gatik

NVIDIA has introduced the Halos AI Systems Inspection Lab, the world's first ANSI National Accreditation Board (ANAB) accredited facility dedicated to inspecting AI-driven physical systems. Announced on March 16, 2026, the lab provides a unified framework for evaluating hardware, software, models, and design principles in AI-based autonomous vehicles (AVs) and robotics. This marks a significant breakthrough in AI safety by integrating functional safety, cybersecurity, AI compliance, and regulatory standards into a single verification environment.

Hesai, a leading lidar provider with over 2 million units delivered globally, has joined the lab to validate its platforms under the ANAB-accredited process. Hesai's ISO 26262 certified products, including those integrated with NVIDIA's DRIVE AGX Hyperion 10 for Level 4-ready vehicles, will undergo rigorous testing. The collaboration, building on a partnership since 2019, aims to embed safety into lidar systems that enable precise 3D environmental perception for hazard detection in AVs and robotics.

Similarly, Gatik, a pioneer in driverless freight, has partnered with the lab to scale its operations across North America. Operating fully autonomous trucks on public roads in Texas, Arkansas, Arizona, and Ontario since 2025, Gatik will use the Halos framework as a real-world proving ground. This includes collaboration with Isuzu Motors for mass-production autonomous platforms and engagement with U.S. Department of Transportation stakeholders.

The lab's full-stack approach unifies safety elements across vehicle architectures and AI models, accelerating rigorous inspection methods and strengthening reliability for large-scale deployments. By making safety an operational discipline, it supports the evolution of standards as autonomous fleets grow to hundreds of vehicles, addressing key barriers to widespread adoption.

This development positions NVIDIA Halos as a cornerstone for safe physical AI, fostering industry-wide standards through partners like Hesai and Gatik. As AI powers more real-world applications, such accredited validation infrastructures represent a critical step toward mitigating risks in autonomous systems.