Unitree H2 Dedicated NVIDIA Jetson Orin NX Computing Board Assembly

Some robots leave the factory with an empty seat where the brain can sit. The Unitree H2 Dedicated NVIDIA Jetson Orin NX Computing Board Assembly is built to fill that seat, and it does so inside the robot's own chassis rather than on a bolted-on shelf. For a team that wants an H2 to see, listen and decide for itself, this board is where that work happens.

This product is available through this page, by quotation.

Background

Unitree and the H2 line

Unitree Robotics is known for legged and humanoid platforms that research labs, universities and integrators around the world use as working tools. Its humanoid line is sold in more than one version, and the sources for this board name three: H2, H2 EDU and H2 Plus. The versions differ in how much computing they bring and how much room they leave for the buyer's own development.

That difference explains why a dedicated computing board assembly exists as its own product. A humanoid body provides motion, balance and sensing. Secondary development, meaning the buyer's own perception, speech, planning or learning software, needs a computer that can run it. The board assembly supplies that computer for the H2 family.

Positioning of the Jetson Orin NX board

This assembly is built around an NVIDIA Jetson Orin NX and is rated at 100 TOPS of performance. It is integrated into the robot chassis, so the compute travels with the robot and does not depend on an external laptop or a tethered workstation. It can be installed in PC3 or PC4.

Within the H2 accessory range it sits beside other compute options. A buyer who prefers an x86 processor can look at the Unitree H2 Dedicated Intel Core i7 Computing Board Assembly. A buyer who wants a different NVIDIA module can look at the Unitree H2 Dedicated NVIDIA Jetson AGX Thor Computing Board Assembly. Each is a separate product with its own page.

Technical Specifications

Specifications as published by the manufacturer; confirmed against the datasheet at quotation.

The manufacturer has not published the dimensions, the weight, the power consumption, the memory, the storage, the interfaces and ports, the operating temperature or the manufacturer part number; they are confirmed at quotation.

What does the Jetson Orin NX board add to an H2?

The plainest answer is a place to run your own software. A humanoid that only walks and follows built-in routines is a demonstration. A humanoid that can process a camera stream, respond to speech or test a new behavior on its own hardware is a development platform. The board provides the computing for that second kind of use, with 100 TOPS of performance available for AI workloads.

The second answer is integration. Because the assembly is integrated into the robot chassis, the computer is part of the machine. Teams that take a robot into a classroom, a demonstration hall or a field trial do not have to carry and power a separate computer or manage a cable run back to it. The robot is one unit.

The third answer is flexibility in placement. The board can be installed in PC3 or PC4. For a buyer planning a configuration, this means the compute is not tied to one position, and the slot choice can be settled with the quotation so that it fits the rest of the build.

How does it relate to the H2 EDU and H2 Basic?

The H2 EDU is the version for which this board matters most directly. Its PC4 slot is delivered empty, and an Intel Core i7 or an NVIDIA Jetson Orin NX can be added there as a paid option. An EDU buyer who wants NVIDIA computing for perception and AI work therefore has a clear path: the Jetson Orin NX board fills the open slot.

The H2 Basic is different. It does not include onboard compute for secondary development and does not support compute-board expansion. A buyer who needs to run their own software on the robot should check this before ordering, because a board of this kind is not an upgrade path for that version.

The sources also name an H2 Plus version, but they do not state how it handles compute expansion. If you own or plan to buy an H2 Plus, tell us when you request a quotation and we will answer the compatibility question.

Who is the Unitree H2 Dedicated NVIDIA Jetson Orin NX Computing Board Assembly for?

University and research labs are the first group. They buy a humanoid to study locomotion, manipulation, perception or human-robot interaction, and they need computing on the robot to run experiments. An H2 EDU with this board in its open slot suits that work.

Educators and training programs are the second group. Teaching robotics today means teaching students to deploy models on embedded hardware, and a Jetson-class module is the kind of platform that is common in that field. Having it inside the robot gives students the full chain from sensor to decision to motion on one machine.

Integrators and product teams are the third group. They evaluate a humanoid for a possible application and need to put their own software stack on it. They should decide early whether they want an NVIDIA module or an Intel processor, and the Intel Core i7 board described above is the sibling to compare. Teams with heavier AI needs may also want to review the Jetson AGX Thor board.

Owners of the H2 Basic are not the audience, since that version does not support compute-board expansion.

What is the Unitree H2 Dedicated NVIDIA Jetson Orin NX Computing Board Assembly used for?

1. Onboard perception. Reading camera and sensor data on the robot itself, so that the robot can interpret what is in front of it.
2. Speech and interaction. Running voice recognition and response software for guide, reception or demonstration roles.
3. Behavior testing. Trying new motion and task behaviors on real hardware instead of only in simulation.
4. Teaching embedded AI. Giving students a Jetson-class platform that travels with the robot into the classroom.
5. Secondary development. Hosting a team's own software stack on the robot for research and prototype work.
6. Application evaluation. Letting an integrator assess whether a humanoid with onboard AI computing fits a planned task.

Pricing and Availability

We do not print prices on this page. Quotation is the way to buy: tell us which H2 version you own or plan to order, and we will prepare an offer for the board assembly. Delivery, customs and support are arranged together with the quotation. You can also browse the H2 Accessories and Humanoid Accessories categories for related parts. The Unitree H2 Dedicated NVIDIA Jetson Orin NX Computing Board Assembly is available on this page.

Frequently Asked Questions

How much performance does the board provide?

The board is rated at 100 TOPS of performance.

Is the board integrated into the robot?

Yes. It is integrated into the robot chassis, so the computing stays inside the machine.

Which slots can it be installed in?

It can be installed in PC3 or PC4. Tell us your configuration and we will settle the slot with you.

Does the H2 EDU come with this board?

No. The PC4 slot of the H2 EDU is delivered empty. An Intel Core i7 or an NVIDIA Jetson Orin NX can be added there as a paid option.

Can I fit it to an H2 Basic?

No. The H2 Basic has no onboard compute for secondary development and does not support compute-board expansion.

Which specifications are missing from this page?

Dimensions, weight, power consumption, memory, storage, interfaces and ports, operating temperature and the part number are listed in the sentence under the specification table.

Where can I buy the Unitree H2 Dedicated NVIDIA Jetson Orin NX Computing Board Assembly and what does it cost?

You can buy it through this page by requesting a quotation. The cost depends on your configuration and is given in the quotation, not printed here.

Summary

An H2 with an empty compute slot is a body waiting for a brain. The Unitree H2 Dedicated NVIDIA Jetson Orin NX Computing Board Assembly supplies that brain: 100 TOPS of performance, integrated into the robot chassis, installable in PC3 or PC4, and a natural fit for the open PC4 slot of the H2 EDU. It does not suit the H2 Basic. If you are planning an H2 that thinks for itself, this page is the starting point.