Reference SoC Platform Focuses on Physical AI Robotic Control

[Ahmet Ö.]

## Reference SoC Platform Focuses on Physical AI Robotic Control

MIPS and Inova Semiconductors have developed an innovative control architecture for humanoid and edge robotic systems, combining a RISC-V processor, regional communication design, and secure edge AI capabilities.

Robotic platforms increasingly require tight integration of computing, connectivity, and control units to support physical AI workloads and real-time motion control. To this end, MIPS, a GlobalFoundries subsidiary, and Inova Semiconductors GmbH have introduced a control reference platform for humanoid and edge robotic applications.

### Importance of Hardware-Based SoC Architecture

The reference platform is designed as a hardware-based system-on-chip (SoC) architecture that supports a complete robotic signal chain, encompassing sensing, processing, actuation, and communication stages. The impact of size, weight, and power constraints on system design in humanoid robots, robotic arms, and other physical AI systems was considered.

The architecture supports mixed-criticality computing, capable of simultaneously running real-time control loops and secure AI workloads. This is crucial for robotic platforms that require specific response times and local AI processing.

The platform was developed using GlobalFoundries’ FDX manufacturing technology with the 22FDX® process, aiming to deliver low power consumption and predictable performance characteristics.

### Convergence of RISC-V Processor and Regional Communication Design

The platform applies regional architecture expertise derived from the automotive data ecosystem to robotic control design. Inova’s high-speed APXpress interface has been integrated with MIPS’s various processing units.

These include: MIPS Atlas M8500 RISC-V high-performance microcontroller IP, MIPS Atlas S8200 RISC-V AI processor IP, and MIPS Atlas mixed-signal technologies. Together, these components enable multi-axis motion control and high-performance data exchange across different network topologies.

The design aims to reduce the software load for data communication by shifting more functions to hardware. This shortens development times and facilitates adaptation to different robot structures.

### Faster Robotic Development with Reference Design

This platform is positioned as a design template for manufacturers developing humanoid and advanced robotic systems. The regional architectural approach supports modular design, enabling the reuse of architectural components across different product variants.

Furthermore, by combining computing and communication functions on a dedicated SoC platform, it aims to reduce bill-of-materials complexity and support secure on-device AI operations alongside critical robotic workloads.

### Simulation Environment Accelerates Software Development

Early access to the platform is provided through the MIPS Atlas Explorer environment. This simulation-based hardware/software co-design platform offers virtual models of the processing architecture, allowing software teams to optimize visual-language-action models and related robotic control software before physical hardware is available.

This co-design approach enables earlier validation of robotic software stacks and AI models compatible with the targeted control architecture.

Inova Semiconductors’ website and GlobalFoundries’ FDX technology form the foundation of this advanced SoC platform.

---

This reference SoC platform stands out as a significant step for manufacturers looking to develop innovative solutions in industrial robotics and AI applications.