Unitree Go2 Docking Station
Unitree Go2 docking station 40 TOPS — external AI computing module for the Go2 EDU, built on NVIDIA Jetson Orin Nano 8GB with a 6-core CPU and 8 GB memory. Identical ports and housing to the 100 TOPS version; best for entry-level vision and teaching workloads.
- NVIDIA Jetson Orin Nano 8GB — 40 TOPS
- Same ports and housing as the 100 TOPS version
- Ubuntu, ROS 2 and JetPack preinstalled
About the Unitree Go2 Docking Station
The Unitree Go2 docking station is an external AI computing module that adds NVIDIA Jetson Orin performance to the Go2 quadruped. It is built in two versions — the 40 TOPS model around a Jetson Orin Nano 8GB, and the 100 TOPS model around a Jetson Orin NX 16GB — and turns the Go2 into a self-contained platform for on-robot perception, mapping and learning.
The two versions of the Unitree Go2 docking station share one housing, one set of ports and one software environment; only the internal compute module differs. The 40 TOPS version suits entry-level computer vision, teaching and lightweight inference, while the 100 TOPS version keeps headroom for advanced SLAM, vision-language models and reinforcement learning. Choosing between them is a question of workload, not connectivity.
Powered by a 16–60 V DC input, the Unitree Go2 docking station brings a 1024-core NVIDIA Ampere GPU with 32 Tensor Cores plus a bank of expansion interfaces — USB, gigabit Ethernet and an M8 aviation connector — so cameras, lidar and other payloads have somewhere to plug in. With Ubuntu, ROS 2 and the NVIDIA JetPack stack on board, the Go2 becomes a mobile development platform that runs inference where the robot walks, not on a desktop back in the lab.
A Compute Module, Not a Charger
Despite the name, the Unitree Go2 docking station does not charge the robot — it is an AI computing module that adds Jetson Orin processing power and extra I/O to the Go2. If you are looking for autonomous charging, that is the Go2 charging station, a different accessory.
Unitree Go2 Docking Station: 40 TOPS vs 100 TOPS
The 40 TOPS and 100 TOPS versions of the Unitree Go2 docking station look identical and share the same 1024-core NVIDIA Ampere GPU architecture with 32 Tensor Cores — what changes is the Jetson module inside, and with it the CPU core count, memory and raw AI throughput.
Pick the version by workload, not by ports: every interface is present on both.
| 40 TOPS | 100 TOPS | |
|---|---|---|
| Compute module | NVIDIA Jetson Orin Nano 8GB | NVIDIA Jetson Orin NX 16GB |
| AI performance | 40 TOPS | 100 TOPS |
| CPU | 6-core ARM Cortex-A78AE @ 1.5 GHz | 8-core ARM Cortex-A78AE @ 2.0 GHz |
| Memory | 8 GB LPDDR5 | 16 GB LPDDR5 |
| Best suited for | Entry-level computer vision, teaching, lightweight inference | Advanced SLAM, vision-language models, reinforcement learning |
What Does the Unitree Go2 Docking Station Add to a Go2?
Offload perception to a dedicated GPU. With an NVIDIA Jetson module at its core, the Unitree Go2 docking station gives the robot a 1024-core Ampere GPU with 32 Tensor Cores — compute headroom for camera inference, mapping and model execution on the move.
Give payloads somewhere to connect. The dock brings a cluster of expansion interfaces — three USB ports, two gigabit Ethernet jacks, a 100 Mbps service port and an M8 aviation connector — so secondary cameras, lidar units and industrial sensors can be wired straight into the robot's compute instead of hanging off a laptop.
Ubuntu, ROS 2 and JetPack on Board
The Unitree Go2 docking station runs the standard Jetson software stack: Ubuntu 20.04 or 22.04, ROS 2, the NVIDIA JetPack SDK with CUDA and TensorRT. For a development team this means the robot arrives as a working environment rather than a bare board — model deployment, sensor drivers and ROS 2 nodes can start on day one.
Which Robots Does the Unitree Go2 Docking Station Fit?
The docking station is made for the Unitree Go2 EDU and Go2 EDU Plus quadrupeds — the variants that carry the robot's expansion interface. It does not fit the wheeled Go2-W, which uses its own dedicated docking station version. Confirm the robot model before ordering.
- Fits: Unitree Go2 EDU and Go2 EDU Plus
- Does not fit: Go2-W — a dedicated Go2-W docking station version exists
Specifications
| 40 TOPS | 100 TOPS | |
|---|---|---|
| Compute module | NVIDIA Jetson Orin Nano 8GB | NVIDIA Jetson Orin NX 16GB |
| AI performance | 40 TOPS | 100 TOPS |
| CPU | 6-core ARM Cortex-A78AE @ 1.5 GHz | 8-core ARM Cortex-A78AE @ 2.0 GHz |
| GPU | 1024-core NVIDIA Ampere, 32 Tensor Cores | 1024-core NVIDIA Ampere, 32 Tensor Cores |
| Memory | 8 GB LPDDR5 | 16 GB LPDDR5 |
| Expansion ports | 1× USB 3.0 Type-A, 1× USB 2.0 Type-C, 1× USB 3.0 Type-C, 2× Gigabit Ethernet (RJ45), 1× 100 Mbps Ethernet (GH1.25-4PIN), 1× M8 aviation connector | 1× USB 3.0 Type-A, 1× USB 2.0 Type-C, 1× USB 3.0 Type-C, 2× Gigabit Ethernet (RJ45), 1× 100 Mbps Ethernet (GH1.25-4PIN), 1× M8 aviation connector |
| Power input | 16–60 V DC | 16–60 V DC |
| Software | Ubuntu 20.04/22.04, ROS 2, JetPack SDK, CUDA, TensorRT | Ubuntu 20.04/22.04, ROS 2, JetPack SDK, CUDA, TensorRT |
| In the box | 1 × Unitree Go2 Docking Station (40 TOPS) | 1 × Unitree Go2 Docking Station (100 TOPS) |
Shipping & Care
The Unitree Go2 docking station ships in a compact, well-padded parcel by express courier. Keep the connector caps in place until installation, mount the unit clear of the robot's leg travel, and route cables so they cannot snag during lying-down or stand-up motions.
Last updated: September 2026.
Frequently Asked Questions
Can't find what you're looking for? Contact our
