Questions about Unitree Go2 and Jetson Orin setup (Section 4.6)

[followingcode]

Hi authors, thanks so much for sharing this excellent work! I'm really interested in the real-world navigation experiments described in Section 4.6 of your paper.

I have a couple of quick questions about the hardware setup:

1. Could you clarify which version of the Unitree Go2 quadruped robot was used? (u1 Standard, u2 Smart, u3 Aurora Smart, or u4 Flagship?)
2. For the Jetson Orin core controller, was it the Jetson Orin Nano or Jetson Orin NX?

[Allenxinn]

Thank you for your interest!

1. We are using the Go2 EDU version, as only this version provides API;
2. The Jetson Orin NX 16G is used in the deployment, but the Jetson Orin Nano is also supported.

[followingcode]

Thank you for your detailed reply! I found that the Unitree Go2 EDU series is divided into four sub-versions: U1, U2, U3, and U4, which differ in computing power and LiDAR configuration. To better reproduce your hardware setup, could you please clarify which exact EDU sub-version (U1/U2/U3/U4) you used in your real-world navigation experiments?

Differences Between Go2 EDU U1, U2, U3, and U4 Versions
All four versions inherit all functions from the Go2-Edu Zero model and are equipped with a RealSense D435i depth camera, plus an optional D1 servo robotic arm. The key differences lie in their computing power and navigation sensors:
Go2edu-U1
Equipped with an expansion module offering 40 TOPS of computing power, including AI algorithms and technical support.
No built-in LiDAR sensor; navigation relies on the depth camera and other basic sensors.
Go2edu-U2
Features a high-performance expansion module with 100 TOPS of computing power, providing stronger AI algorithm support and technical services.
No built-in LiDAR sensor, same as U1.
Go2edu-U3
Also has the 100 TOPS high-computing expansion module.
Adds a Mid-360 3D LiDAR from Livox, including navigation algorithms and technical support for advanced SLAM and autonomous navigation.
Go2edu-U4
Retains the 100 TOPS high-computing expansion module.
Upgrades to a premium XT16 3D LiDAR from Hesai, with more precise ranging and wider field of view, paired with professional navigation algorithms and support.

[Allenxinn]

We are using the EDU-U2. The only difference between the U1 and U2 lies whether the compute module is an Orin Nano or an Orin NX, or you can also purchase an official NV jetson module and install it yourself.

[followingcode]

That's exactly right! The key difference lies in the TOPS performance. The 40 TOPS version corresponds to the Jetson Orin Nano 8GB, while the impressive 100 TOPS high-performance version matches the Jetson Orin NX 16GB. On the other hand，we employ the Unitree GO2 as our quadruped robot platform, which transmits the real-time ego-centric video stream to a server equipped with a single RTX4090 GPU via a video transmission protocol. What protocol is used for the video transmission?