Sensing, powering, and safeguarding: how Murata components support the technology foundation of AI and humanoid robots

As humanoid robots move from laboratories to factories, hospitals, and homes, and as AI servers carry increasingly massive model inference tasks, a fundamental question emerges: the implementation of intelligence must ultimately be underpinned by the reliability of hardware. For a humanoid robot to walk steadily, grasp dexterously, and collaborate safely with humans, countless electronic components must work together with precision on a millisecond timescale. For AI servers to continuously deliver computing power, they likewise cannot do without the guarantees of stable power supply and thermal management. This article will introduce the product characteristics of Murata's inertial measurement units (IMUs), millimeter-wave radar, soft-termination multilayer ceramic capacitors (MLCCs), thermistors, and power-over-coax (PoC) inductors, and sort out the functions and features of these products in AI and humanoid robots, attempting to present a picture of hardware support "from perception to execution, from signal to power."

In the field of humanoid robots, Murata's IMU (inertial measurement unit) mainly plays the core role of balance sensing. By precisely measuring the robot's posture, motion, and orientation, it helps the robot remain stable and achieve autonomous navigation in complex environments.
 
The IMU is the foundational sensor for humanoid robots to achieve high-precision motion control. Its functions cover multiple levels from local stability to global navigation, such as being applied in the head for perception fusion and visual stabilization, cooperating with cameras, LiDAR, and others to provide a stable posture and positioning reference when visual sensors are disturbed or fail (such as during rapid movement or in poor lighting), ensuring the continuity of SLAM (simultaneous localization and mapping). When applied at the hip/torso, it can perform real-time posture and gait control, monitoring body tilt angle and center-of-gravity changes in real time at a high frame rate, providing key feedback for the balance control of bipedal walking, and ensuring the robot can cope with uneven ground or external disturbances.
 
In addition, the IMU can also be applied to the arms/hands for fine motion tracking, capturing small posture changes of the arms and fingers, and assisting the force control system in achieving more precise grasping, placement, and other operations. When applied to the humanoid robot as a whole, it can perform autonomous navigation and positioning. In indoor environments without GPS, it estimates position through inertial navigation algorithms, supporting the robot's autonomous movement in complex terrain.
 
Represented by the SCH1633-D05, Murata's IMU design precisely responds to the stringent requirements of humanoid robots for high precision, high robustness, small size, and easy integration. The sensor is factory-calibrated to achieve high precision and simplified integration, and due to the extremely limited installation space in humanoid robots, on-site calibration after installation is very difficult. The SCH1633-D05 adopts extended factory calibration technology, characterizing and compensating the sensor behavior across the full operating temperature range (-40 to 110°C) before shipment, which means customers no longer need to perform cumbersome and expensive secondary temperature calibration. At the same time, it completes axis orthogonality correction at the factory, simplifying the development process for system integrators, and its gyroscope total offset performance is below 0.15^∘/s, ensuring long-term stability.
 
In addition, Murata's IMU also features high robustness and functional safety design. Facing the vibration, impact, and severe temperature changes of industrial and outdoor scenarios, sensors are required to be extremely reliable. The SCH1633-D05 is based on a mature 3D MEMS process and has high shock resistance. More critically, it has more than 200 internal monitoring signals built in and supports the ISO26262 functional safety standard, enabling powerful fault detection, which is crucial for safety-critical robot applications.
 
Murata's IMU also has the characteristics of miniaturization and integration friendliness. In order to adapt to the extremely limited space inside robot joints, the SCH1633-D05 adopts a low-profile plastic SOIC package with integrated EMC shielding, which not only saves space but also effectively resists electromagnetic interference from processors and communication modules, which is very critical in robot systems handling multi-sensor fusion, and supports multi-sensor fusion and synchronization. Robots need to fuse multiple sensor data such as vision, IMU, and torque. The SCH1633-D05 supports the SafeSPI V2.0 interface, with built-in SYNC input, data-ready output, and timestamp functions, making it easy to achieve clock domain synchronization with cameras and main control chips, significantly reducing the development difficulty of multi-sensor fusion systems.

In the fields of AI and humanoid robots, Murata's soft-termination MLCCs (represented by the GCJ series) mainly play the role of guarantors of high-reliability power and signal links. Their core value lies in the fact that when robot joints perform intense, high-frequency motion, ordinary capacitors may crack and fail due to circuit board bending, while soft-termination MLCCs ensure stable power supply and system safety through their unique structural design.
 
The working environments of humanoid robots and AI servers place extremely high demands on the mechanical stress of electronic components. The function of soft-termination MLCCs is precisely to solve these pain points, such as resisting mechanical stress and preventing fracture, because when humanoid robot joints move, the circuit board is subjected to continuous bending, vibration, and impact. The ceramic body of traditional MLCCs is brittle and prone to cracks as a result, leading to short circuits or open circuits. Soft-termination MLCCs add a conductive resin layer to the external electrodes, which can effectively absorb and release the stress generated by circuit board deformation, thereby greatly reducing the risk of ceramic body cracking. This is crucial for key locations such as joint servo drive boards.
 
In addition, soft-termination MLCCs can also stabilize AI computation and power supply. AI chips (such as GPU/NPU) have huge power consumption fluctuations during operation and require capacitors to be placed close to the chip for instantaneous power supply (decoupling) to stabilize voltage. While providing high capacitance, soft-termination MLCCs' high reliability ensures that power filtering and voltage regulation functions will not fail due to mechanical stress during robot movement, thereby safeguarding the real-time performance and stability of AI inference and motion control.
 
Taking Murata's GCJ series as an example, its product characteristics embody Murata's comprehensive strength in materials, design, and processes. Its core innovation lies in the resin electrode's ability to absorb stress. The most notable feature of the GCJ series is its "soft termination" structure. It does not directly plate metal on the ceramic body, but instead introduces a resin layer into the external electrodes. When the PCB bends, the resin layer deforms first, transferring stress away from the fragile ceramic body, thereby "suppressing cracks caused by board bending." This technology greatly improves mechanical reliability without sacrificing capacitance performance.
 
Murata's GCJ series also features miniaturization and high capacitance. In order to adapt to the compact space inside robots, Murata continues to achieve higher capacitance in smaller packages. For example, its GCJ series launched the world's first soft-termination MLCC to achieve 2.2µF/100Vdc in an 0805-inch (2.0 x 1.25mm) package. Compared with the previously required larger 1206-inch products, its mounting area is reduced by about 51%, saving valuable board space for designers. In addition, Murata's GCJ series also supports complete application coverage, providing a rich product line covering multiple package sizes from 0603-1210 for low-voltage and 0805-2220 for high voltage, as well as different rated voltages such as 25V, 50V, and 100V, and offering multiple temperature characteristics such as X7R, X7T, and X7S to meet the needs of different locations from AI computing modules to joint drive boards.

In the fields of AI servers and humanoid robots, Murata's thermistors (NTC and PTC) play the role of "thermal management safety guards." They are respectively responsible for precise temperature monitoring and active overcurrent/overheat protection, ensuring that high-computing-power chips and precision motors operate within a safe temperature range.
 
NTC (negative temperature coefficient) thermistors are mainly used for high-precision temperature monitoring. The resistance of NTC thermistors decreases as temperature rises, and they are mainly used for temperature detection and compensation. NTC thermistors can be used in AI servers, closely attached to major heat sources such as GPUs and CPUs, monitoring temperature in real time, and providing key data for fan speed control and system frequency reduction to prevent overheating crashes. In humanoid robot applications, they can monitor the temperature of motor windings, drive boards, and batteries, preventing joint motors from burning out due to overheating and ensuring motion safety.
 
Murata's NTC thermistors have high precision and high reliability, with resistance tolerance as low as ±0.5% and B-value tolerance up to ±0.5%, ensuring accurate temperature measurement, and excellent thermal response, using small SMD packages (such as 0402, 0603), capable of quickly sensing temperature changes and providing timely feedback, and with a rich product line, offering multiple choices from consumer grade to automotive grade (such as the NCU series), with operating temperatures up to 150°C, adapting to harsh environments.
 
PTC (positive temperature coefficient) thermistors (POSISTOR) can support active overcurrent and overheat protection. The resistance of PTC thermistors increases sharply after exceeding a specific temperature (Curie point), and they are mainly used for overcurrent protection and overheat detection. Their core function is circuit protection. They can be connected in series in power or motor drive circuits. When abnormal current increases and causes self-heating beyond the Curie point, the resistance surges, suitability for overcurrent protection depends on the specific product rating and application conditions. In addition, PTC thermistors also support overheat detection. When mounted near heat sources such as power transistors (FETs) and power ICs, when the temperature exceeds a set threshold, the resistance changes sharply, triggering the protection circuit.
 
Murata's PTC thermistors have the characteristics of fast response and precise protection. By virtue of the sharp change in resistance, they have strong noise immunity and can use simple circuits to achieve precise overheat detection, reducing the occupation of MCU ports. They also have high reliability. Based on ceramic devices, they have a solid-state structure, resistant to mechanical vibration and impact, suitable for vibration environments such as robot joints. They also offer flexible threshold settings, providing multiple Curie point options (such as 50°C to 130°C), which can be flexibly matched according to the tolerable temperature of different motors or power devices.

In the fields of AI and humanoid robots, Murata's 60 GHz millimeter-wave radar modules mainly play the role of "all-weather environmental perception safety redundancy." Using the characteristic that millimeter waves are resistant to lighting changes and more robust in challenging weather conditions than vision-based sensing, they can complement cameras and LiDAR under harsh operating environments, providing robots with reliable presence detection, obstacle avoidance, and safety protection capabilities.
 
The value of millimeter-wave radar in humanoid robots is concentrated in its ability to provide the system with a perception dimension independent of vision, such as compensating for visual blind spots and achieving all-weather obstacle avoidance. Because camera performance can degrade significantly in low-light glare, or high-contrast environments, and depth perception is also prone to errors when facing transparent or highly reflective surfaces such as glass and mirrors, creating situations where additional sensing modalities can improve reliability. 60 GHz radar can often detect obstacles such as glass doors or reflective surfaces more reliably than vision-only systems, providing complementary environmental information for safe navigation.
 
In addition, millimeter-wave radar can also perform human presence detection and millimeter waves can penetrate clothing and fabrics and detect micro-movements such as breathing and heartbeat, capable of performing coarse motion or activity sensing. This makes it very suitable for presence detection, sensing even when a person is stationary. For humanoid robots, this means it can more reliably determine a person's presence, location, and movement characteristics. Studies have already demonstrated real-time teleoperation of a humanoid robot arm using only a single millimeter-wave radar. On the other hand, millimeter-wave radar can also serve as safety redundancy to meet functional safety requirements. In functional safety design, sensors based on different principles serving as redundancy for each other is crucial.
 
Taking Murata's Type1VN and Type2LP 60 GHz radar modules as examples, they are highly integrated system-in-package solutions. Their feature lies in the integration of TI's chipset, antenna, and peripheral circuits, reducing the design threshold for customers.
 
The Type1VN is very compact in size, suitable for robot joint or head areas with strict space requirements. Its integrated DSP and MCU can directly process radar signals inside the module and output preliminarily processed target information (such as distance, speed, and angle), reducing the computing burden on the robot's main control.
 
The current mainstream direction is not to replace cameras with radar, but to fuse them. The cooperation between TI and NVIDIA is an example. Fusing millimeter-wave radar data with camera data and connecting it to robot computing platforms such as NVIDIA Jetson Thor can significantly improve the accuracy of target detection and tracking while reducing false alarms and achieving more reliable real-time decision-making. Murata's modules, as high-performance radar front ends, can be easily integrated into such fusion perception architectures.

In the fields of AI and humanoid robots, Murata's PoC (power over coax) inductors mainly play the role of "intelligent separators for signal and power highways." Their core value lies in ensuring that when a robot's vision sensor transmits high-definition video signals and DC power simultaneously through a single coaxial cable, the two do not interfere with each other, thereby simplifying wiring, reducing weight, and improving reliability.
 
The principle of PoC technology is to superimpose high-frequency data signals and DC power on a single coaxial cable. PoC inductors (also called Bias-T inductors) form a "Bias-T circuit" in the circuit, whose role is to block high-frequency signals from flowing into the power line while allowing DC power to pass smoothly.
 
In humanoid robots, this technology directly solves several key pain points, such as simplifying joint and head wiring. A robot's vision sensors (such as cameras and LiDAR) are usually installed in the head or torso, while the main control board is in the center of the body. Traditional solutions require one signal line plus one power line. PoC technology requires only one coaxial cable to complete both power supply and communication, greatly reducing the number of wire harnesses passing through joints, making wiring simpler and more lightweight.
 
In addition, PoC technology can ensure the integrity of high-definition video signals. Robots need to process high-definition video streams in real time for AI inference and navigation. PoC inductors must maintain high impedance over a very wide frequency band to prevent high-speed signals (such as GMSL2 up to 6 Gbps) from leaking into the power line, thereby avoiding signal attenuation or waveform distortion and ensuring that the AI vision system obtains clear, interference-free image data.
 
On the other hand, PoC technology can also improve system reliability. Fewer cables and connectors reduces the number of interconnect failure points. PoC simplifies the system architecture and reduces potential mechanical failure points, which is crucial for humanoid robots that need to move frequently.
 
Murata's PoC inductors, especially its LQW32FT series, are designed specifically to meet the above severe requirements. Their product characteristics directly respond to the challenges in robot design, such as supporting broadband high impedance, with a single inductor replacing multiple inductors. Traditional Bias-T circuits often require multiple inductors with different inductance values in series to maintain high impedance over a wide frequency band. Murata's LQW32FT series achieves high impedance characteristics over a wide band through self-developed ceramic materials and winding structures, allowing a single inductor to replace the multiple devices previously required. This not only saves valuable PCB space but also reduces total DC resistance (DCR).
 
Murata's PoC inductors also support large current and adapt to compact spaces. The power consumption of humanoid robot vision modules is not low, and the space is extremely limited. The LQW32FT_8H model in this series achieves a rated current of up to 1.6A (such as the LQW32FT1R6M8H model) in a small 1210-inch (3.2 x 2.5mm) size, meeting the power supply needs of high-power cameras in narrow spaces. In addition, Murata's PoC inductors have a wide operating temperature range of -40°C to +125°C, which supports operation across a wide range of thermal environments.

The value of the Murata components discussed in this article in AI and humanoid robots can be summarized in three keywords: perception, power supply, and safeguarding. The common characteristic of these components is that they do not directly constitute the core of AI algorithms or robot motion control, but they are the prerequisite for the stable operation of these core functions. As humanoid robots pursue higher degrees of freedom, more compact structures, and longer endurance, the reliability, integration, and energy efficiency of components will increasingly become bottlenecks and breakthrough points in system design. Through continuous iteration in materials, processes, and packaging technologies, Murata is providing a reliable hardware foundation for the AI and robotics industries.