Humanoid Robot Chips: What STMicroelectronics Watches Before Choosing MCUs or MPUs | MyTechDailyNews General

Humanoid robots are still moving from laboratory gestures toward real jobs, and the chip question underneath that shift is not fully answered. STMicroelectronics is watching the market rather than betting immediately on a bigger MCU or MPU route, according to Allan Lagasca, who leads the company’s robotics segment strategic program.

For readers following the sector, the debate centers on where a robot should think: near the hand, foot, or camera that senses an object, or inside a central processor that keeps the body coordinated. STMicroelectronics says fast reaction should stay local, while whole-body coordination belongs to a GPU. The company has not picked a more powerful MCU or MPU path because demand is still low and designs have not stabilized.

“Every millisecond counts,” Lagasca told EE Times. “If you want a fast reaction, you should move the processing to the edge. If you want the entire system to work together, that is where the GPU comes in.”

Allan Lagasca

Why Local Processing Can Reduce Robot Lag

A round trip to a central GPU can add delay that a moving robot cannot afford. Lagasca said placing inference nearer to the task lowers the workload sent to a central GPU and avoids delays tied to moving data out and returning commands. In robotics, those delays can show up as physical errors.

If two hands coordinate to transfer an object, a late command can make a joint miss its target. Even misses of only a few millimeters can affect the combined movement. “If inference is performed at the edge, the system can send the necessary command locally and respond faster,” Lagasca said.

The Central GPU Is a “Larger Brain,” Not a Replacement for Local Control

Lagasca described the central GPU as the robot’s “larger brain,” but it still must synchronize head cameras, machine vision and navigation, and coordinate joints whose sensors are not connected to one another. Without coordination, he said, “the robot could move both feet forward at the wrong time.” STMicroelectronics does not make GPUs.

The company’s current AI inference products are based on its neural processing unit (NPU) and microprocessor unit (MPU). “By following this market, we will better understand which technologies are needed,” Lagasca said. Robotics remains a low-volume market, while STMicroelectronics, as an integrated device manufacturer, targets high-volume production.

Hand-Level Intelligence: A Gesture Path From Camera to Motor

A robotic hand demonstration at electronica India used a camera to capture an image of a human hand, after which an AI model identified 21 points to recognize the gesture. That information was then translated into commands for a robotic hand with six motors. The gesture recognition ran on STMicroelectronics’ STM32N6 32-bit microcontroller.

“The mapping is not simple,” Lagasca said. “It is not even a one-to-one or one-to-two relationship, and it must run continuously as the hand moves.”

The work starts with the camera, whose module combines an image sensor, an inertial measurement unit (IMU) to determine the camera’s position and orientation, and a time-of-flight (ToF) sensor to measure distance. “Capturing an image is not enough; you also need to measure the distance,” Lagasca said. The ToF reading tells the hand when to close its fingers.

Lagasca said inference also runs mainly onboard the camera, partly for privacy. As long as an image is used for inference without being stored or sent elsewhere, the system can meet privacy and cybersecurity requirements, he added.

The camera’s output must then become commands the motors can use. For the gesture demonstration, Lagasca said a standard PLC bridged AI outputs and robotic control. “The difficulty lies in translating the data generated by AI into a format that can be transferred to a standard PLC to control the robot,” he said.

That edge approach also lets the hand adjust its grip. A human adjusts a hold on a water bottle almost instinctively; a robot has to compute the required grip. Lagasca gave another example: a hand that identifies a water bottle, determines its distance, and adjusts its grip to avoid crushing it or letting it slip. “The MCU, in this case the STM32, performs this inference,” he said.

Each part of the robot handles its share. Lagasca said most sensors have their own MCUs, as do motion and motor-control functions, while AI inference has its own NPU. “This creates a distributed architecture rather than putting everything into a single processor,” he said.

Module Integration and Hardware Simplification

Because humanoid robot manufacturers may source hands, joints, and cameras from different suppliers, integration can create problems with camera placement, board space, and motor control. A camera mounted on top of a hand may require the robot to reposition its hand before lifting an object, while one mounted underneath may strike the table during pickup.

STMicroelectronics is exploring ways to integrate camera technology and other functions into the humanoid hand itself. Although the company does not manufacture camera modules, it can combine its technologies with those of partners. Earlier this year, STMicroelectronics and camera-module maker Leopard Imaging introduced a multimodal vision module for humanoid robots that combines STMicroelectronics’ imaging technology, 3D depth-mapping, and motion sensors. The module connects over Ethernet through Nvidia’s Holoscan Sensor Bridge to Nvidia’s Jetson computers and Isaac robot development platform, and it is designed to fit the size, weight, and power limits of humanoid robots.

The broader approach, Lagasca said, is to divide the robot into modules, including hands, joints, feet, and eyes, that can be adapted to different applications and form factors.

Packaging those functions inside modules is another challenge. Designers must fit processing, motor control, and connectivity into the available board space. Lagasca cited a customer requirement for a 6 × 6-centimeter board.

He said choosing MCUs, motor drivers, and connectivity parts separately could produce a design measuring 140 × 330 mm, far larger than the board. For its robotic hand, ST integrated six MOSFETs and three drivers into one package, so a manufacturer can mount one part instead of nine.

Open Safety Questions

Because intelligence is distributed across a robot, safety concerns also arise. Lagasca said existing safety frameworks include Safety Integrity Level (SIL) requirements for industrial applications and Automotive Safety Integrity Level (ASIL) requirements in the automotive sector. These requirements may become relevant as companies from the automotive, industrial automation, smartphone, and home appliance markets enter robotics.

Compliance with a standard, however, does not guarantee that the entire robot is safe. “If a sensor fails, the robot could still cause an injury, even if it meets the relevant SIL or ASIL standard,” Lagasca said.

He said AI can introduce additional risks if a system fails to make a decision or makes the wrong one. All elements must be considered together to assess functional safety. Lagasca added that many existing standards were developed for fixed equipment rather than moving robots, and the industry still needs to determine which standards and regulatory requirements are appropriate for humanoid systems, particularly those that interact directly with people.

Chip Opportunity in the Humanoid BOM

Lagasca said STMicroelectronics can address approximately 600 positions in a humanoid robot’s bill of materials (BOM), covering more than 500 components. He added that the share of a robot’s total cost represented by those parts depends on design, including how many joints, sensors, and AI functions the robot has.

Illustration	BOM value	STMicroelectronics parts value	Share
Full-supply case	$600	$600	100%
Larger BOM case	$1,000	$600	60%

Entry-level robots may cost less and have fewer degrees of freedom, while mid-range and high-end systems may need more sophisticated capabilities. Lagasca said that, depending on design, STMicroelectronics could potentially address 800 or more positions as joints, sensors, cameras, and AI-processing requirements increase.

From Demos to Deployment

Lagasca said many applications remain at the proof-of-concept stage despite more humanoid robot demonstrations. Automotive manufacturing and logistics are early areas of interest because manufacturers can justify automation cost and adjust production lines to meet market demand.

As examples of links between humanoid robotics and the automotive industry, he cited Boston Dynamics’ association with Hyundai and Figure AI’s relationship with BMW. Hyundai, which owns Boston Dynamics, plans to start deploying Atlas robots at its Georgia plant in 2028, while BMW has conducted trials for Figure robots in its Spartanburg, S.C., body shop.

Moving beyond demonstrations has several challenges. Training data is one. Lagasca said activities such as drilling, soldering, cooking, and sewing can be recorded and uploaded to create datasets. Battery technology does not yet support 24-hour operation, making downtime a practical consideration for developers.

Applications that reach the market also depend on safety and regulatory requirements. Lagasca said direct physical interaction between humanoid robots and people presents additional challenges, particularly in healthcare. Cybersecurity is another concern: chipmakers and device manufacturers must account for the EU Cyber Resilience Act and other applicable requirements at both product and application levels. “You do not want 10,000 robots operating in the real world only for someone to hack into them,” he said.

A Practical Watchlist for Non-Technical Readers

Watch where the processing lives: local edge chips handle fast reactions, while a central GPU coordinates the whole body.
Watch hand demonstrations such as the electronica India example, because gesture control shows how 21 points, six motors, and a 32-bit STM32N6 path can work together.
Watch module integration, since camera placement, board space, and motor control can break humanoid robot designs.
Watch safety conversations around SIL, ASIL, AI decision errors, failed sensors, and standards for moving robots.
Watch early commercial fields: automotive manufacturing and logistics, including Hyundai, Boston Dynamics, Atlas, BMW, Figure AI, and Figure.
Watch practical barriers: training data, battery limits that do not yet support 24-hour operation, healthcare interaction, and cybersecurity rules such as the EU Cyber Resilience Act.

Common Misunderstandings

A central GPU does not need to handle every reflex; a moving robot cannot afford a slow round trip.
A humanoid robot is not one big processor; it is closer to a distributed system with sensors, motors, cameras, and inference each doing part of the job.
Meeting a safety standard does not automatically make a whole robot safe, because a failed sensor can still cause injury.
STMicroelectronics is not deciding only from chip capability; it is watching demand, volume, architecture choices, and market needs.
More joints, sensors, cameras, and AI functions can raise the addressable BOM, but the percentage depends on the final design.

Action Tips

For product teams: ask whether a task needs local edge inference or whole-body coordination before choosing MCU or MPU resources.
For readers evaluating robot news: separate demonstration from deployment and check whether training data, battery life, safety, regulation, and cybersecurity are addressed.
For market watchers: track module suppliers, integration limits, and humanoid BOM complexity instead of headline robot counts.

FAQ

Why is STMicroelectronics waiting?

Humanoid designs remain unsettled and volumes are low, so STMicroelectronics is observing the market before committing to a product path, according to Allan Lagasca. “By following this market, we will better understand which technologies are needed,” he said.

What does the edge versus GPU debate mean?

Does STMicroelectronics make the central GPU?

No. STMicroelectronics does not make GPUs, while it positions its neural processing unit and microprocessor unit products for inference tasks.

How big is the parts-list opportunity?

Lagasca said STMicroelectronics can address approximately 600 positions in a humanoid robot’s BOM, covering more than 500 components, and could potentially address 800 or more positions in more complex designs.

Still Searching for the Right Shape

How intelligence is ultimately split between edge processors and the central GPU may depend on an unsettled design question: what a humanoid robot should look like. The system architecture “has not stabilized,” Lagasca said. “Some developers ask why a robot should have only two hands rather than six, or only two legs.”

“We do not yet know what the final configuration will be,” Lagasca said. For everyday readers, that uncertainty is the story: humanoid robots are becoming more tangible, but the chip path is still waiting for the market shape to clarify.