Application Case | Anchoring Innovation, JHCTECH Humanoid Robot Control Solution

2024 is also known as the first year of humanoid robot commercialization. Humanoid robots are the result of a deep integration of artificial intelligence and robotics technologies. They can freely enter and exit various environments without being limited by scenarios or regions and are equipped with brains capable of thinking. Humanoid robots integrate advanced technologies such as artificial intelligence, high-end manufacturing, and new materials, and are expected to become a disruptive innovation following computers, smartphones, and new energy vehicles, leading to a leap in productivity.

Humanoid Robots

A humanoid robot is a type of robot that utilizes artificial intelligence and robotics technology to resemble human appearance and behavior. These robots achieve the simulation of human-like functions through various sensing, transmission, computing, and control components. They can autonomously analyze their surroundings, automatically generate and execute tasks, and interact with humans through voice communication. Additionally, they can collaborate with other robots and smart devices such as industrial robots, AGVs, and robotic vacuum cleaners through various sensing equipment, playing an important role in a wide range of production and daily life scenarios.

The humanoid robot industry chain can be divided into three key aspects: the brain (perception, decision-making, and human-computer interaction), the cerebellum (motion control) and the limbs. Among them, the controller serves as both the brain and cerebellum of the humanoid robot, forming the core foundation of industry development.

Humanoid robot controller

Compared to other mature industries, the principles of humanoid robot controllers are fundamentally similar, but there are differences in specific structures and requirements.

• Industrial Robots:The controller is used to control the trajectory and spatial position of the robotic arm. In comparison to humanoid robots, industrial robots demand higher control precision and a deeper understanding of processes.
• Robotic Vacuum Cleaners: The controller primarily handles path planning, obstacle avoidance, and human-machine interaction, which are also functions of a humanoid robot controller. However, the real-time algorithm requirements and processing capabilities for humanoid robots are much higher.
• Automotive Domain Controllers:Typically divided into powertrain, chassis, cockpit, autonomous driving, and body domains, the control principles of autonomous driving domain controllers are similar to those of humanoid robots. Both require multi-sensor fusion, positioning, path planning, decision-making, image recognition, high-speed communication, and data processing. However, there are differences in safety requirements, control precision, algorithm complexity, and interface complexity between automotive and humanoid robot controllers.

Requirements for humanoid robot controllers:

⭐ As a highly intelligent AI device, a humanoid robot requires extremely high overall performance from its main controller due to its need for high computing power, real-time processing, synchronization, and complexity.

⭐ Humanoid robots achieve advanced coordination of hand, eye, brain, and body in human-machine interaction. Their motion control involves the synchronized operation of nearly a hundred joints and sensors, demanding that the controller respond quickly and accurately manage the actions of each component.

⭐ The controller must continuously adjust operation plans and trajectories in real-time, coordinating the state of both legs and the entire body. This balance is achieved through the collaboration of various types of sensors.

⭐ To improve the robot’s autonomous capabilities, the controller must support deep learning of various data samples. The control architecture, which involves visual perception, motion control, and interaction algorithms, is highly complex and requires immense computational power to handle the large volume of data.

JHCTECH Solution

The BRAV-7131 AI edge computing controller is developed and designed based on NVIDIA Jetson AGX Orin 64GB, with an AI computing power of up to 275 TOPS. Its efficient AI performance meets the requirements of algorithms such as spatial fusion and multimodality, enabling it to complete a large number of high-performance, high-precision target recognition, image processing and analysis tasks in a short time. This greatly enhances the perception and decision-making capabilities of humanoid robot controllers.

It has a rich IO interface to meet the functional requirements of humanoid robots: multiple LAN ports support peripherals such as servo control and machine vision; independent CAN communication satisfies the communication needs for high reliability, such as automatic charging; and multiple USB 3.0 and above interfaces meet the needs of high-speed transmission, effectively completing the control and interaction of humanoid robots.

NVIDIA Jetson AGX Orin provides a wealth of development tools and resources to meet algorithm requirements such as spatial fusion, multimodality, navigation and obstacle avoidance, path planning, motion mapping, handling, interaction, and decoupling. It equips the BRAV-7131 with leading AI computing and motion control capabilities, enabling developers to create high-performance humanoid robot products more quickly and efficiently. Additionally, it helps companies shorten R&D cycles and reduce costs, enhancing the competitiveness of JHCTECH’s AI edge computing controllers in the market and ensuring optimal solutions for humanoid robot control.

BRAV-7131

• NVIDIA Jetson AGX Orin 32/64G, 200/275TOPS
• 8/12*Cores ARM Cortex-A78AE CPU, 2.2GHz(maximum frequency)
• NVIDIA Ampere GPU with Tensor Cores
• Onboard 32/64G 256bit LPDDR5 Ram and 64GB eMMC，1*M.2 M-Key NVME and optional 1*2.5“SATA bay
• 1*HDMI, 1*Audio Line-out，5*LAN, 2*USB3.2, 2*USB2.0, 4*COM, 2*CAN, 16-bit DIO,1*SYNC IO
• 1*F-Mini PCIe and 1*M.2 B-Key support 4G LTE or 5G NR，1*M.2 E-Key support WiFi& BT wireless module
• DC9~36V wide power input, with OVP, OCP, and SCP
• Aluminum-magnesium alloy chassis, Fan cooling design

In addition, JHCTECH is about to release the new BRAV-7121, developed based on NVIDIA Jetson NX Orin, with a maximum computing power of up to 100 TOPS. With its efficient AI capabilities and compact design, it can be widely applied in intelligent robotics. Let's take a look at it in advance!

The BRAV-7121 is developed based on the NVIDIA Jetson NX Orin, offering computing power of 70/100 TOPS and a low power consumption of just 25W. It features a 6/8-core ARM Cortex-A78AE CPU and an Ampere architecture GPU, onboard 8/16GB of memory, and supports external NVMe storage. The device features a variety of IO interfaces, and supports a wide DC power supply range of 9-36V, making it ideal for applications in intelligent robotics, smart transportation, machine vision, and smart logistics.

• NVIDIA Jetson Orin NX 8/16G,70/100TOPS
• 6/8*Cores ARM Cortex-A78AE v8.2 64 bit CPU
• Ampere GPU（1024 cores），32*Tensor cores
• 8/16G LPDDR5，1*M.2 2280 M-Key NVMe
• 2*LAN（POE optional ），2*Iso.CAN（One of the CAN channels with isolation is optional），4*GMSL(optional)；
• Video code:1x4k60 | 3x4k30 | 12x1080p30 H265
• Video decode:1x8k30 | 4x4k30 | 18x1080p30 H265
• 1*HDMI,1*Line-out ,2*USB3.2,,2*USB2.0, 1*Debug，2*Iso.COM,1*8-bit Iso.DIO
• 1*M.2 3052 B-Key+SIM ; 1*MiniPCIe；
• DC-IN 9~36V wide power input ,  DC-OUT 12V;
• Aluminum alloy chassis, Fanless cooling design

As the next AI wave, humanoid robots are expected to create significant demand for controllers as they become industrialized in the future. While the controller is more “intangible” than the actuator, it plays a crucial role in the performance of the robot itself. JHCTECH will continue to innovate in research and development, empower product innovation with scenarios, and support the growth and implementation of the humanoid robot industry.

Learn more about BRAV-7131 series

Learn more about BRAV-7121 series