Summary
This knowledge article explains how STMicroelectronics MEMS sensors are empowering robotics and physical AI. Two main topics are presented: the application tree available on the ST website and the collaboration with NVIDIA.
Introduction
Physical AI is the next frontier of artificial intelligence and refers to the integration of artificial intelligence in physical systems, where sensors, actuators, and computing are key components.
The STMicroelectronics portfolio contains more than 500 components, including microcontrollers, drivers, sensors, and modules. These components cover different market segments: industrial, humanoid, medical, and service robotics.
In robotics applications, there is strong interest in humanoid robots, which combine multiple technologies and are designed to work safely and effectively with people in dynamic environments. Some examples of these applications are shown at CES 2026:
Humanoid robot companies can use STMicroelectronics components in different units. Sensors are present in the body sensing unit, hand unit, head unit, and articulations (legs and arms).
On the robotics application page of the st.com website, there is a dedicated page for humanoid robots that explores the different units.
For example, in the Articulations (legs and arms) block diagram for humanoid robots, you can check the suggested devices in the Sensing block. This includes IMUs and other types of sensors.
1. Get started with ST evaluation and development tools
Tools and software are also suggested for the Sensing block of the diagram. One of the recommended boards for industrial robotics applications is the expansion board X-NUCLEO-IKS5A1, which enables development with STM32 Nucleo boards.
The X-NUCLEO-IKS5A1 board can be connected to various STM32 Nucleo boards.
This configuration reduces development effort and accelerates time to market.
Use the combined boards with the software included in the X-CUBE-MEMS1 package and MEMS Studio.
STMicroelectronics sensors offer advanced features, including low-power sensor fusion (SFLP), finite state machine (FSM), machine learning core (MLC), and intelligent sensor processing unit (ISPU). For more information, see the page "MEMS sensors ecosystem for machine learning".
2. Collaboration with NVIDIA and sensor integration in Isaac Sim
STMicroelectronics accelerates the adoption of physical AI with NVIDIA by integrating sensors, microcontrollers, and motor control solutions with NVIDIA Holoscan Sensor Bridge and Isaac ecosystems. This helps developers design, train, and deploy humanoid robots with higher efficiency, reliability, and scalability.
Isaac Sim provides a controlled virtual world where motion can be generated repeatedly and safely. Robot trajectories can be simulated by attaching virtual IMU sensors. The ideal motion can be studied in simulation, and the tool helps translate movements into inertial measurements.
Ideal simulation alone is not enough. Real IMUs have noise, bias drift, vibration effects, temperature sensitivity, and device-specific behavior. The Sim2Real IMU model adds the missing sensor realism needed to bridge simulation and reality.
In simple terms, Isaac Sim provides the robotic motion environment, while the Sim2Real IMU model makes the inertial sensor behavior realistic. Together, this creates a practical workflow for generating synthetic IMU data, validating algorithms, and accelerating motion intelligence development before deployment on real systems.
After selecting the sensor model in the NVIDIA Isaac Sim software, when the robot executes trajectories in the simulation environment, the sensor output is immediately usable for downstream pipelines. Logging, analytics, and external streaming are some of the features available in the tool.
Robotics and perception teams can evaluate software stacks against more realistic inertial sensor behavior without requiring custom scripting each time a sensor is added to a scene.
Once developers specify how they expect the robot to behave, they can deploy their policies to an NVIDIA Jetson Thor™ or NVIDIA Jetson Orin™. This allows developers to efficiently benefit from Holoscan Sensor Bridge by streaming information to the computing unit.
Conclusion
With STMicroelectronics MEMS sensors, you can develop your robotic application efficiently. An ecosystem of tools (including boards and software) is provided to facilitate sensor evaluation and application development.
Sensor support in other ecosystems, such as NVIDIA Holoscan Sensor Bridge and Isaac Sim, further help to speed up development.
Create your solutions and advance robotics to the next level. Stay tuned on the latest innovations in STMicroelectronics MEMS sensors for robotics, and watch the dedicated webinar "From motion to intelligence: MEMS sensors in robotics and physical AI".
Related links
