Advantech Advances Physical AI and Robotics With Integrated Functional Safety

TAIPEI, Taiwan - Oct 22, 2026 - Advantech (TWSE: 2395), a global leader in industrial IoT, has expanded its Physical AI and robotics platform to address the growing requirements for safe, validated, and deployable intelligent robots. As AI expands into the physical world, robots are evolving into Physical AI systems capable of perceiving their surroundings, processing information, making decisions, and interacting with physical environments. To address the requirements of robot development, Advantech presents a five-layer Robot Platform architecture comprising layer1 Perception, layer2 AI Computing & Motion Control, layer3 Functional Safety, layer4 Production & Validation, and layer 5 Deployment & Management. The architecture integrates key capabilities across the robot lifecycle, from perception and AI computing to safety, validation, and deployment.

"To scale robotics solutions in real environments, it requires safety to be considered in every layer, from the ground up. We’re closing the gap with NVIDIA IGX Thor, offering an industrial grade platform that’s purpose-built for safety-certifiable robotics applications. " said Magic Pao, Vice President of Applied Server Group, Advantech.

Building a Complete Robot Platform

Developing robots for real-world deployment requires the integration of computing, sensing, motion control, safety, software, and validation. Advantech brings these capabilities together through an established ecosystem of platforms, technologies, and solution partners, helping Robot OEMs reduce the complexity of system development and integration. Advantech’s platform integrates key components such as cameras and other sensing technologies, reduces the effort and time required for component integration and system testing. With safety-oriented design, diagnostic mechanisms, and supporting safety documentation, including safety analysis and FMEDA-related evidence, Advantech helps customers reduce the effort required for system-level safety integration and validation. In addition, NVIDIA IGX Thor is part of the NVIDIA Halos safety platform for autonomous robots, helping developers engineer safety across design, validation and deployment. Advantech platforms support NVIDIA software and technologies, enabling customers to connect physical systems with simulation and digital environments for development and validation. Together, these capabilities help robot OEMs accelerate development, reduce integration and validation effort, and move more efficiently toward real-world deployment.

Functional Safety for Physical AI

As robots move beyond controlled environments and increasingly interact with people, machines, and dynamic physical environments, safety must become an integral part of the system architecture alongside perception and AI performance. 

Functional Safety (FuSa) is therefore an increasingly important capability within the Physical AI platform. Advantech integrates safety architecture, Safety MCU, diagnostic mechanisms, Safety Manual, FMEDA, and engineering support into its robot platform. Its approach separates the AI compute path from an independent safety path, enabling the AI roadmap to support AI inference, vision, and control while the independent safety path monitors system health, detects faults, and provides mechanisms to support application-defined safe-state handling. 

Advantech’s MIC-735-IT, powered by NVIDIA IGX Thor, can be configured with a Renesas Safety MCU and supports safety-related system integration. Advantech’s FuSa-ready approach provides the platform design, diagnostic mechanisms, and supporting evidence needed to support the customer’s functional safety lifecycle. Through DMS and system integration capabilities, Advantech can evaluate application-specific requirements across hardware, firmware, Safety MCU, and external safety controllers. 

Advantech provides platform-level design and integration support to help customers address applicable functional safety standards, including IEC 61508 and ISO 13849. Final hazard analysis, system integration, validation, and system certification are carried out by the system integrator or end customer based on the specific application requirements, with Advantech providing platform-level design and integration support throughout the development process.

Advantech Showcases Safe Physical AI at NVIDIA GTC Berlin 2026

At NVIDIA GTC Berlin 2026, Advantech will demonstrate how AI computing, perception, and functional safety are integrated into Physical AI systems. 

The MIC-742-AT will demonstrate multi-sensor synchronization, real-time AI computing, and robotic perception. The MIC-735-IT showcase a FuSa-oriented reference architecture based on NVIDIA IGX Thor and NVIDIA Halos, demonstrating the separation of the AI Compute Path and Independent Safety Path and how system-level diagnostic and safety mechanisms can be integrated into robotic platforms. 

Through these demonstrations, Advantech will present its five-layer Robot Platform spanning Perception, AI Computing & Motion Control, Functional Safety, Production & Validation, and Deployment & Management, providing a structured approach to the development and deployment of Physical AI systems. 

Advantech provides system-level capabilities for the next generation of Physical AI and intelligent robotics. Learn more about Advantech’s Functional Safety Solution: Advantech FuSa Solution.