STMicroelectronics Integrates Robotics Hardware with NVIDIA AI Platforms

[Cengiz Özemli]

## STMicroelectronics Integrates Robotic Hardware with NVIDIA AI Platforms

STMicroelectronics is taking a significant step in the development, training, and deployment of humanoid and industrial robotic systems by integrating its sensors, microcontrollers, and motor control solutions with NVIDIA's AI-powered robotics platforms.

### New Integration in Robotics and Physical AI Systems

Humanoid robots and industrial automation systems require tight integration of sensor data, control, and simulation environments to accelerate their development processes. In this context, STMicroelectronics is expanding its collaboration with NVIDIA to incorporate its sensor and control solutions into NVIDIA's robotics ecosystem. This aims to enable faster and more scalable development of physical AI systems.

### NVIDIA Holoscan Sensor Bridge Integration

A key part of the collaboration is the integration of ST components with the NVIDIA Holoscan Sensor Bridge (HSB). HSB enables standardized acquisition, synchronization, and processing of sensor and actuator data. This allows developers to:

• Connect multiple sensors and actuators through a single interface
• Simplify data collection and logging processes
• Create consistent datasets for AI models

This approach is crucial for robotic systems that require the synchronized use of various sensing technologies, such as inertial measurement units (IMUs), image sensors, and time-of-flight (ToF) devices.

### From Simulation to Reality with High-Fidelity Simulation Models

Another important aspect of the partnership is the integration of high-fidelity digital models of ST components into the NVIDIA Isaac Sim simulation environment. The first model is an IMU, followed by ToF sensors and other integrated circuits. Derived from real hardware data, these models closely mimic real device behavior in simulations.

This results in:

• Reduced discrepancies between simulation and the real world
• Improved convergence of AI training models
• Decreased inefficiencies caused by the use of random parameters

More accurate simulations lead to shorter development times and reduced costs associated with prototype testing.

### Ease of Hardware and Software Co-Design

STMicroelectronics and NVIDIA are also simplifying the integration of hardware with AI platforms. STM32 microcontrollers, sensors, and motor control systems are being made compatible with NVIDIA Jetson platforms. This allows:

• AI models to be trained in simulation
• Sensor and actuator behaviors to be accurately represented
• Minimization of adjustments during the transition to physical deployment

This integration optimizes sensing, motion, and control coordination, especially in humanoid robots and complex autonomous systems.

### Solution for Complexity and Scalability in Robotics

Advanced robotic systems require high computational power, large datasets, and precise simulation settings. Incorrect modeling and high parameter variability can lead to training inefficiency and a decline in real-world performance. STMicroelectronics and NVIDIA aim to mitigate these challenges by offering hardware-validated models and standard integration pathways.

### Leadership in the Robotics Ecosystem

This collaboration is part of the trend of integrating semiconductor technologies with AI development platforms. By combining its sensor, microcontroller, and motor control portfolio with NVIDIA's AI and simulation infrastructure, STMicroelectronics is providing a unified development environment for next-generation robotic applications.