NVIDIA Halos：为何规模化部署物理 AI 需要全栈安全

物理人工智能（Physical AI）：https://www.nvidia.com/en-us/glossary/generative-physical-ai/ 正在从研究迅速向大规模部署发展。根据 ABI Research：https://my.abiresearch.com/research/15976/ 的预测，到 2035 年，3-5 级自动驾驶车辆（AVs）的装机量将达到 4900 万辆，而 Omdia：https://omdia.tech.informa.com/om146251/robotics-hardware-market-forecast--2026 估计在 2026 年至 2035 年间将部署大约 6000 万台工业机器人。随着这些机器进入道路、工厂、仓库及其他人类共享的环境，安全性必须与之同步提升。

物理人工智能安全意味着需要证明 AI 驱动的机器——自动驾驶车辆（AVs）：https://www.nvidia.com/en-us/glossary/autonomous-vehicles/、类人机器人（humanoid robots）：https://www.nvidia.com/en-us/glossary/humanoid-robot/、工业机器人等——在其决策转化为物理行动时行为是安全的。这要求在硬件、软件、AI、操作环境和部署生命周期各方面确保安全——而不仅仅是在部署前进行一次性检查。

经过多年的测试和基准评测，自动驾驶车辆（AVs）：https://www.nvidia.com/en-us/solutions/autonomous-vehicles/ 继续在商业上扩张。这个进展要求开发者展示自动化系统如何应对潜在的硬件和软件故障、功能限制以及 AI 特有风险。

随着自动化机器进入工厂、仓库及其他人类共享的环境，机器人技术正接近类似的转折点。

在物理人工智能领域，制造商、监管机构、保险公司和工作场所安全团队需要证据，证明硬件、软件、AI 行为和操作环境能够协同工作并在无需人工干预的情况下保持安全。

四种转变定义了新的安全标准：

这些转变共同要求在设计、部署和验证全过程中实现安全，从底层硬件到 AI 行为再到操作环境。

物理人工智能安全需要专业的工程、数据、流程和验证，少数公司单独无法完全复制。NVIDIA 的安全基础源于十多年来在自动驾驶安全领域的开发经验，积累了功能安全、传感器融合、AI 行为保证、视觉 AI、仿真及现实验证等方面的专业知识。

NVIDIA Halos：https://www.nvidia.com/en-us/ai-trust-center/halos/autonomous-vehicles/?deeplink=use-case-tabs--2 是第一个也是唯一一个用于物理 AI 的全栈安全系统，帮助开发者在设计、验证和部署的每一层面进行安全工程。其原则在自动驾驶汽车（AV）和机器人中共享，而平台、标准和证据仍各自特定于各个领域。

这些要素将基于云的 AI 开发和仿真与车内部署连接起来，使安全证据可以在车辆生命周期中保持可追溯。

在 AV 和机器人领域，NVIDIA Halos AI 系统检查实验室：https://www.nvidia.com/en-us/ai-trust-center/physical-ai/safety-certification/ 将安全、网络安全和 AI 安全要求转化为可重复的检查，并帮助准备 Halos 集成以供第三方机构进行最终系统级认证。

NVIDIA Halos 连接了构建、集成、评估和部署物理 AI 解决方案的公司，包括产品开发者、软件和嵌入式系统提供商、传感器和芯片公司、安全解决方案开发者以及认证机构。

在自动驾驶汽车领域，吉利、五十铃、日产（由 Wayve 软件驱动）和 Einride 正在基于 NVIDIA Hyperion 构建 L4 级别准备就绪的车辆，并由 Halos OS 提供支持。

Uber、Grab、Lyft 及其他出行服务提供商也在使用 Hyperion 扩展 Robotaxi 的开发和部署。NVIDIA Halos AI 系统检查实验室的成员：https://www.nvidia.com/en-us/ai-trust-center/physical-ai/safety-certification/?_gl=1*qks3dj*_gcl_aw*R0NMLjE3ODc3NTcxODguQ2owS0NRanduYnJVQmhET0FSSXNBS0toUHBleEl2bnlBMGloYWR2bUEyTlIxVjlycFVSaVViZnZuYnFySWZOaUZaUGp3bnlnU3RfNTIxVWFBdTI1RUFMd193Y0I.*_gcl_au*MTM2NDIwNjE5Mi4xNzg4Mzk0NDg1Li0uLS4xNzg4Mzk0NTQ0LjEyNjE4OTg4MzkuMTc4OTE0OTcwMy4xNzg5MTczNzgz 包括 AUMOVIO、博世、Gatik、Hesai、Lucid、MIRA、onsemi、PlusAI、索尼、法雷奥和 Wayve，涵盖自动驾驶开发、ADAS、传感器、芯片、系统集成、验证和安全保障。

在机器人领域，acontis 和 QNX 提供运行安全功能所需的嵌入式软件，而 Advantech（https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.advantech.com%2Fen%2Fresources%2Fnews%2Fadvantech-mic-735-brings-functional-safety-to-physical-ai-systems&data=05%7C02%7Cpfox%40nvidia.com%7C0223fcf382c746a60e5d08df1201f825%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C639249471820682901%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=jQ3sOLBihRlPDCJXaYi34Po9HVvyhUAnAByBPhB0TTQ%3D&reserved=0）和 NexCOBOT 构建了安全设计的 NVIDIA IGX 系统。Infineon、NXP、STMicroelectronics 和 Texas Instruments 提供传感器、安全微控制器及其他半导体技术。KION 集团正在开发用于自主叉车的功能安全代理。Agility 正在将 NVIDIA IGX Thor 和 Halos Core 集成到其 Digit 5 仿人机器人（https://www.agilityrobotics.com/content/agility-unveils-digit-5-humanoid-robot-built-for-cooperatively-safe-work-at-scale）的安全系统中。

对于自动驾驶汽车，TÜV SÜD 已认证 NVIDIA 的汽车产品生命周期软件流程和 DriveOS 6.0 符合 ISO 26262 ASIL D 标准，以及 NVIDIA 的汽车工程流程符合 ISO/SAE 21434 标准。TÜV Rheinland 还对 NVIDIA DRIVE AV 进行了独立的 UNECE 安全评估。

在机器人领域，TÜV Rheinland 正在检查 NVIDIA IGX Thor、Halos OS 和 Holoscan 传感器桥的功能安全认证准备情况，这建立在 TÜV SÜD 对 Thor SoC 和 Halos Core 的 ISO 26262 检查基础上。

在物理 AI 领域，ANAB 已将 NVIDIA Halos AI 系统检测实验室认证为 ISO/IEC 17020 检测机构。该实验室检查范围内的 Halos 集成，并帮助公司为由独立第三方机构进行的最终认证做好准备。

那些推进物理 AI 规模化的公司，不仅仅会构建最强大的系统。他们将构建可以被评估、认证、部署并在现实世界中被信赖的系统。从一开始就设计功能安全是原型与可扩展解决方案的关键区别。

了解更多关于自动驾驶汽车的 NVIDIA Halos：http://nvidia.com/en-us/ai-trust-center/halos/autonomous-vehicles/ 以及机器人技术：https://www.nvidia.com/en-us/ai-trust-center/halos/robotics/，并探索面向物理人工智能的全栈安全架构。

Physical AI ：https://www.nvidia.com/en-us/glossary/generative-physical-ai/ is moving rapidly from research to large-scale deployment. By 2035, ABI Research ：https://my.abiresearch.com/research/15976/ projects an installed base of 49 million level 3-5 autonomous vehicles (AVs) , while Omdia ：https://omdia.tech.informa.com/om146251/robotics-hardware-market-forecast--2026 estimates that roughly 60 million industrial robots will be deployed between 2026 and 2035 . As these machines enter roads, factories, warehouses and other environments shared with people, safety must scale with them.

Physical AI safety means proving that AI-driven machines — AVs ：https://www.nvidia.com/en-us/glossary/autonomous-vehicles/ , humanoid robots ：https://www.nvidia.com/en-us/glossary/humanoid-robot/ , industrial robots and more — behave safely when their decisions turn into physical action. That requires safety across the hardware, software, AI, operating environment and deployment lifecycle — not a one-time check before deployment.

After years of testing and benchmarking, AVs ：https://www.nvidia.com/en-us/solutions/autonomous-vehicles/ continue to expand commercially. That progress has required developers to demonstrate how automated systems address potential hardware and software failures, limitations in intended functionality and AI-specific risks.

Robotics is approaching a similar inflection point as autonomous machines move into factories, warehouses and other environments shared with people.

Across physical AI, manufacturers, regulators, insurers and workplace safety teams need evidence that hardware, software, AI behavior and operating environments can work together safely without human intervention.

Four shifts define new safety standards:

Together, these shifts require safety to be operationalized across design, deployment and validation, from the underlying hardware to AI behavior and the operating environment.

Physical AI safety requires specialized engineering, data, processes and validation that few companies can reproduce alone. NVIDIA’s safety foundation draws on more than a decade of development in AV safety, building expertise in functional safety, sensor fusion, AI behavior assurance, vision AI, simulation and real-world validation.

NVIDIA Halos ：https://www.nvidia.com/en-us/ai-trust-center/halos/autonomous-vehicles/?deeplink=use-case-tabs--2 is the first and only full-stack safety system for physical AI, helping developers engineer safety across every layer of design, validation and deployment. The principles are shared across AVs and robotics, while the platforms, standards and evidence remain specific to each domain.

Together, these elements connect cloud-based AI development and simulation with in-vehicle deployment so safety evidence can remain traceable across the vehicle lifecycle.

Across both AV and robotics, the NVIDIA Halos AI Systems Inspection Lab ：https://www.nvidia.com/en-us/ai-trust-center/physical-ai/safety-certification/ turns safety, cybersecurity and AI safety requirements into repeatable inspections and helps prepare Halos integrations for final system-level certification by third-party agencies.

NVIDIA Halos connects the companies that build, integrate, assess and deploy physical AI solutions, including product developers, software and embedded-system providers, sensor and silicon companies, safety solution developers and certification bodies.

In autonomous vehicles, Geely, Isuzu , Nissan (powered by Wayve software) and Einride are building level 4-ready vehicles on NVIDIA Hyperion, supported by Halos OS.

Uber, Grab, Lyft and other mobility providers are also using Hyperion to scale robotaxi development and deployment. Members of the NVIDIA Halos AI Systems Inspection Lab ：https://www.nvidia.com/en-us/ai-trust-center/physical-ai/safety-certification/?_gl=1*qks3dj*_gcl_aw*R0NMLjE3ODc3NTcxODguQ2owS0NRanduYnJVQmhET0FSSXNBS0toUHBleEl2bnlBMGloYWR2bUEyTlIxVjlycFVSaVViZnZuYnFySWZOaUZaUGp3bnlnU3RfNTIxVWFBdTI1RUFMd193Y0I.*_gcl_au*MTM2NDIwNjE5Mi4xNzg4Mzk0NDg1Li0uLS4xNzg4Mzk0NTQ0LjEyNjE4OTg4MzkuMTc4OTE0OTcwMy4xNzg5MTczNzgz include AUMOVIO, Bosch , Gatik , Hesai, Lucid , MIRA, onsemi , PlusAI, Sony, Valeo and Wayve, spanning autonomous-driving development, ADAS, sensors, silicon, systems integration, validation and safety assurance.

In robotics, acontis and QNX provide the embedded software needed to run safety functions predictably, while Advantech ：https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.advantech.com%2Fen%2Fresources%2Fnews%2Fadvantech-mic-735-brings-functional-safety-to-physical-ai-systems&data=05%7C02%7Cpfox%40nvidia.com%7C0223fcf382c746a60e5d08df1201f825%7C43083d15727340c1b7db39efd9ccc17a%7C0%7C0%7C639249471820682901%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=jQ3sOLBihRlPDCJXaYi34Po9HVvyhUAnAByBPhB0TTQ%3D&reserved=0 and NexCOBOT build safety-designed NVIDIA IGX systems. Infineon , NXP , STMicroelectronics and Texas Instruments contribute sensor, safety-microcontroller and other semiconductor technologies. KION Group is developing functional safety agents for autonomous forklifts. Agilit y is integrating NVIDIA IGX Thor and Halos Core into the safety system for its Digit 5 humanoid ：https://www.agilityrobotics.com/content/agility-unveils-digit-5-humanoid-robot-built-for-cooperatively-safe-work-at-scale .

For AVs, TÜV SÜD certified NVIDIA’s Automotive Product Lifecycle software process and DriveOS 6.0 to ISO 26262 ASIL D, as well as NVIDIA’s automotive engineering processes to ISO/SAE 21434. TÜV Rheinland also performed an independent UNECE safety assessment of NVIDIA DRIVE AV.

For robotics, TÜV Rheinland is inspecting NVIDIA IGX Thor, Halos OS and Holoscan Sensor Bridge for functional-safety certification readiness, building on TÜV SÜD’s inspection of the Thor SoC and Halos Core for ISO 26262.

Across physical AI, ANAB has accredited the NVIDIA Halos AI Systems Inspection Lab as an ISO/IEC 17020 inspection body. The lab inspects scoped Halos integrations and helps companies prepare for final certification by independent third-party bodies.

The companies that scale physical AI will not simply build the most capable systems. They will build systems that can be assessed, certified, deployed and trusted in the real world. Designing functional safety from the start is what separates a prototype from a scalable solution.

Learn more about NVIDIA Halos for AVs ：http://nvidia.com/en-us/ai-trust-center/halos/autonomous-vehicles/ and robotics ：https://www.nvidia.com/en-us/ai-trust-center/halos/robotics/ , and explore the full-stack safety architecture for physical AI.