Unitree Go2 Docking Station 100 Tops

A quadruped can walk almost anywhere, yet it is only as clever as the computer that reads its sensors. The Unitree Go2 Docking Station 100 Tops is the official expansion module that gives the Go2 a serious AI computer of its own, built around an NVIDIA Jetson Orin NX with 16GB of LPDDR5 memory. For teams that want perception and autonomy work to happen on the robot instead of on a laptop at the edge of the room, it is the piece that makes the difference.

The product is available through this page, by quotation.

Background

Unitree Robotics and the Go2 lineage

Unitree Robotics is known for its quadruped robots, which research labs, universities and industrial teams use as mobile platforms. The Go2 is one of the best known of those platforms, and around it the company has built a family of accessories: batteries, controllers, charging hardware, protective parts and expansion modules. The docking station belongs to that family. It is an official expansion module designed for the Unitree Go2, so it is made for this robot and not adapted from a generic computing box.

That matters for anyone who has tried to bolt an outside computer onto a robot. Cables, power supplies and software setups all become your own project. An official module narrows that work, because the robot maker has defined the target platform from the start.

Positioning and key technology

The docking station is an AI Computing Docking Station. Its core is the NVIDIA Jetson Orin NX with 16GB of LPDDR5 memory, delivering 100 TOPS of AI performance. Alongside the compute module sits an 8-core ARM Cortex-A78AE processor running at 2.0GHz and a 1024-core NVIDIA Ampere architecture GPU running at 918MHz, with 32 Tensor Cores.

The software side follows the NVIDIA ecosystem that robotics developers already know: Ubuntu 20.04 and 22.04, ROS 2, the NVIDIA JetPack SDK, CUDA and TensorRT. In practical terms, a team that has trained a model with common tools can think about deploying it with TensorRT, and a team that builds on ROS 2 can keep its existing structure.

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 operating temperature, the IP rating, the storage capacity or the wireless connectivity; they are confirmed at quotation.

What makes the 100 TOPS version different from the 40 TOPS configuration?

The docking station comes in two configurations. The one on this page uses the NVIDIA Jetson Orin NX with 16GB of memory and is rated at 100 TOPS. The other uses the smaller NVIDIA Orin Nano with 8GB of memory and is rated at 40 TOPS. It is sold separately and is better suited to educational and entry-level development.

The difference is about headroom. A single perception model may run comfortably on either module, but real robot projects rarely stay with one model. Teams add object detection, then mapping, then a planning layer, and each one competes for the same memory and processing. The larger module gives those workloads more room to run side by side. If your work is a classroom exercise or a first experiment with onboard AI, the smaller configuration may be all you need. If you expect to stack several tasks, or to keep growing the software over a long project, the 100 TOPS version is the safer starting point.

How does the docking station connect to cameras, LiDAR and other devices?

Through its ports. The module offers one USB 3.0 Type-A port and two USB 3.0 Type-C ports, two Gigabit Ethernet connections on RJ45, one Fast Ethernet connection on a GH1.25-4PIN connector, and one M8 aviation plug interface. The available USB, Ethernet and aviation interfaces support the integration of cameras, LiDAR and other compatible devices.

Having both USB and Ethernet is useful in practice. USB suits cameras and small peripherals, while wired Ethernet is the usual choice for sensors that stream a lot of data or for a link to another computer on the network. The wide power input of 16V to 60V DC also gives the integrator some freedom when planning how the module is powered.

Who is the Unitree Go2 Docking Station 100 Tops for?

The first group is research laboratories and university groups that already work with a Go2 and want to run perception, navigation or learning-based control on the robot itself. The Jetson software stack means students and engineers can build on tools they are likely to have met before.

The second group is system integrators and development companies that are building a demonstration or a pilot for a customer. They need a compute platform with room to grow, standard interfaces for sensors and a documented software environment. The 100 TOPS version suits them better than the smaller one, because the demands of a pilot tend to rise as the project goes on.

The third group is educators and entry-level developers. For them, the 40 TOPS configuration with the Orin Nano and 8GB of memory is the better fit, and it is available separately. Teams that are still outfitting their Go2 can also look through the Accessories category, the Quadruped Accessories category, or related items such as the Unitree Go2 Self-Charging Board and the Unitree Go2 Servo Robotic Arm.

What is the Unitree Go2 Docking Station 100 Tops used for?

1. Onboard AI perception. Run vision and detection models on the robot rather than on a remote computer.
2. ROS 2 development. Build and test robotics software in a Linux environment with Ubuntu 20.04 or 22.04.
3. Sensor integration. Connect cameras, LiDAR and other compatible devices through the USB, Ethernet and aviation interfaces.
4. GPU-accelerated computing. Use CUDA and TensorRT on the Ampere architecture GPU for inference workloads.
5. Research and teaching. Give students and researchers a capable computing platform on a familiar quadruped.
6. Prototype and pilot projects. Evaluate autonomy ideas on a real robot before committing to a larger deployment.

Pricing and Availability

Pricing for the Unitree Go2 Docking Station 100 Tops is provided by quotation, so that the configuration, quantity and destination are settled first. Delivery, customs and support are arranged with the quotation. Send us your request and the details of your Go2 project, and the product is available on this page.

Frequently Asked Questions

Which robot does this docking station work with?

It is an official expansion module designed for the Unitree Go2.

What AI hardware does it contain?

It is built around an NVIDIA Jetson Orin NX with 16GB of LPDDR5 memory, delivering 100 TOPS. It also has an 8-core ARM Cortex-A78AE processor and a 1024-core Ampere GPU with 32 Tensor Cores.

Which software does it support?

The software environment is Ubuntu 20.04 and 22.04, ROS 2, the NVIDIA JetPack SDK, CUDA and TensorRT.

What ports does it offer?

There is one USB 3.0 Type-A port, two USB 3.0 Type-C ports, two Gigabit Ethernet ports on RJ45, one Fast Ethernet connection on GH1.25-4PIN and one M8 aviation plug interface.

Is there a smaller version?

Yes. A 40 TOPS configuration with the NVIDIA Orin Nano and 8GB of memory is available separately. It is better suited to educational and entry-level development.

What about the weight and the IP rating?

Those figures, along with the dimensions and the operating temperature, are not listed on this page. Ask for them with your quotation request.

Where can I buy the Unitree Go2 Docking Station 100 Tops and what does it cost?

You can request it through this page. The cost is given in a quotation that also covers delivery, customs and support, and no price is listed here.

Summary

A walking robot is only half of an autonomous system, and the other half is the computer that makes sense of what the robot sees. The Unitree Go2 Docking Station 100 Tops supplies that half as an official module for the Go2, with an NVIDIA Jetson Orin NX, 16GB of memory, 100 TOPS of AI performance and a familiar software environment. Its USB, Ethernet and aviation interfaces leave room for cameras, LiDAR and other devices. If you want your Go2 to do its thinking on board, this page is the starting point.