Source: GigaDevice
    From the Spring Festival Gala stage to the World Robot Conference, humanoid robots are no longer puppets on the stage but are beginning to transition into true productivity tools. 2025 is a crucial year for China's humanoid robots to move towards large-scale mass production and commercial application, with many leading enterprises announcing significant progress one after another. According to public data, UBTECH's Walkers series has received over 500 orders from car manufacturers and secured a huge order worth 250 million yuan.     Zhiyuan Robotics plans to start large-scale shipments in the second half of the year and is expected to produce several thousand units by 2026. Yushu Technology aims to produce over 5,000 units annually and has launched a lightweight version priced at only 39,900 yuan.
The maturity of the hardware system is precisely the crucial foundation for humanoid robots to achieve large-scale mass production and commercial application. Its environmental perception, motion control and task execution capabilities rely on the efficient coordination of the six core units: the perception layer, rotational joints, linear joints, the overall trunk, dexterous hands and robotic feet. In this architecture, the performance of the chip and efficient communication directly determine the real-time response speed and intelligence level of the robot.

    In the race of chip-enabled robot hardware, GigaDevice, with years of technological accumulation, has built a complete product matrix covering control, storage and simulation, providing full-stack chip support for humanoid robots. Among them, the GD32 high-performance MCU can be used for real-time motion control and system scheduling of multiple joints. The Flash under its umbrella provides high-speed and highly reliable data storage guarantees for decision-making computing. Analog chips cover key links such as sensor signal conditioning, motor drive, and power management, empowering the robot system throughout the entire chain from perception, decision-making to execution.

Improve the matrix to empower precise control
    The motion control system is like the cerebellum of a humanoid robot, responsible for coordinating the driver to control the actuator motor to complete precise movements based on the target trajectory and load status. In this system, the MCU serves as the nerve endings of the main processor, distributed at each joint, undertaking instruction reception, motor drive, encoder /IMU data processing, and closed-loop control of torque, position, and speed. This distributed architecture effectively reduces the load on the main control, ensuring low latency, high precision and multi-joint coordination of movements, and is the core foundation for achieving flexible and stable movements.
    The space of the dexterous hand is limited and the motors are dense, so there is an urgent need for a highly integrated MCU. The GD32G5 series stands out with its ability to drive multiple motors on a single chip. Take the GD32G553 as an example. It adopts the Arm Cortex-M33 core, with a main frequency as high as 216MHz. It integrates a DSP instruction set, a single-precision FPU and a hardware trigonometric function accelerator (TMU), providing strong computing power. It can easily achieve precise control of a single motor or drive 3 to 4 motors simultaneously. Meanwhile, its WLCSP81 package size is only 4×4mm, which greatly saves space. Three chips can cover all the joints of the dexterous hand, significantly optimizing the layout and cost.
    There are numerous joints in humanoid robots, each with different requirements for drive. For instance, the legs are responsible for support and dynamic balance, and need to calculate gait in real time and adjust posture, which demands higher computing power. The GD32H7 series is thus highly suitable. It is equipped with a 600MHz Cortex-M7 core, dual-precision FPU, TMU and filter algorithm accelerator (FAC), and integrates industrial interfaces such as CAN-FD and Ethernet. It CAN complete multi-sensor data fusion on a single chip and is widely used in humanoid robot leg joints and robot dogs.
    The EtherCAT bus is widely used in humanoid robots to integrate communication between joints, sensors and controllers. If this protocol is adopted, the GD32H75E is an ideal choice. This chip integrates the officially authorized EtherCAT IP on the basis of GD32H7, and is equipped with two 100-megabit Ethernet PHyS. It supports multi-axis synchronization at the μs level, meeting the requirements of high-speed and low-latency communication in complex motion scenarios. It can be used in arm joints that need to perform fine grasping and other tasks and have high control accuracy requirements. In addition to its extremely high integration, the GD32H75E also supports 105℃ ambient temperature operation and features a BGA144 (8×8mm) small package, addressing the pain points of limited joint space and high operating temperature in robots.
    The communication node is responsible for protocol parsing and instruction forwarding between robot components. The GD32C113 series can meet the requirements. Adopting the Cortex-M4 core (up to 120MHz), this series also supports 2 high-speed CAN-FDs. If the EtherCAT protocol is used, a 2/3 port GDSCN832 slave controller can be selected. It integrates 2 internal PHyS, 1 MII expansion interface, and is equipped with 8KB dual-port memory, providing big data processing capabilities for complex control.
    Overall, GigaDevice's GD32 MCU can work in conjunction with the GD30DR series driver chips, along with LDO, DC-DC, PWM and protection circuits, to build a complete and reliable robot motor control solution, fully serving joint drive and servo systems.
In conjunction with the GD32 MCU, GigaDevice's storage products also play a key role in humanoid robots. Among them, the GD25/55X /LX series SPI NOR Flash has a data throughput of 400MB/s, which can meet the requirements of immediate startup and real-time response, and provide high-speed storage support for the robot brain that performs AI decisions.

High performance and high integration, breaking through traditional bottlenecks
    In humanoid robot systems, the role of simulation products is equally crucial. Gigadevice has also established a comprehensive simulation product matrix covering core demands, with multiple blockbuster products providing support for the precise control and stable operation of robots.
    The GD30AD3380 is a high-performance ADC with 8-channel 16-bit synchronous sampling, capable of achieving a maximum sampling rate of 1MSPS. It supports true bipolar hardware and software modes, allowing for independent input range configuration for each channel and data transmission through flexible interfaces such as SPI and parallel bus. Its synchronous sampling feature is suitable for high-precision perception requirements and can be widely applied to the data conversion of six-dimensional torque sensors in robot exoskeletons and dexteric hands, enabling precise capture of force and torque information of joints and load-bearing structures by the motion system. Combined with relevant algorithms, it can achieve smooth and delicate motion control.
    The GD30DR1488 is a gate driver specifically designed for three-phase brushless DC motors (BLDC) of robot joints. It has a high voltage tolerance of 150V and is compatible with the operating voltage range of the motor. It provides A pull-up current of +1.0/-1.3A and a maximum switching frequency of 500kHz to ensure the drive response speed. Its core advantage lies in the built-in bootstrap diode, which simplifies the peripheral design and reduces the PCB area.
     GD30DRE518 is a SoC chip with more integrated advantages. This SoC adopts an MCU based on the Cortex-M33 core and integrates a three-phase BLDC pre-drive with a maximum withstand voltage of 150V. The maximum main frequency of its MCU is 180MHz, equipped with 512KB flash memory and 128KB SRAM, and provides 16-channel 12-bit ADC and 3-channel 12-bit DAC. It simultaneously provides multi-channel advanced and conventional general-purpose timers, as well as rich interface resources, including 3-channel CAN-FD,      SPI/I2C/USART/GPIO, etc. It can simultaneously undertake the tasks of node communication, motor drive and sensor data processing, breaking the limitation of traditional motor drive SoC chips that can only drive motors and cannot handle other complex tasks.
    Compared with traditional simulation solutions, GigaDevice's simulation products have significant advantages: they not only break through the single function, achieve miniaturization with high integration, but also can meet various strict requirements. Delving into the competitiveness of its analog product line, it stems from three core supports: First, it relies on product lines such as MCUS, technological accumulation and talent teams to achieve technological collaborative reuse; Second, relying on the customer ecosystem accumulated over the years, we focus on advantageous industries such as robotics, create multi-product line collaborative solutions, and precisely match the demands of various scenarios. Third, relying on continuous strategic investment, combined with a high-quality supply chain and a first-class quality system, we provide customers with reliable and cost-effective products.

Conclusion

    The year 2025 is regarded as a crucial turning point for humanoid robots to move towards large-scale mass production, and it is expected to drive a significant increase in the demand for high-precision motor control, multi-modal sensing and edge computing chips. To address this trend, GigaDevice will focus on areas such as real-time control of servo drives, a complete portfolio of analog products, and high-performance storage products, providing customers with cost-effective one-stop solutions to help humanoid robots achieve breakthroughs in core technologies and large-scale applications.

    免责声明： 本文章转自其它平台，并不代表本站观点及立场。若有侵权或异议，请联系我们删除。谢谢！ 

  Disclaimer: This article is reproduced from other platforms and does not represent the views or positions of this website. If there is any infringement or objection, please contact us to delete it. thank you!
    矽源特科技ChipSourceTek