[Unitree Go1] Sharing internet with NVIDIA Jetson [ip 13] for installing repo packages

Yes, your ROS multimachine setup is incorrect.

I meant to say your multi machine setup from the images are incorrect, The one we’ve written is correct and is applicable for when a pc that is connected to the go1 via LAN cable having 192.168.123.1 as the static ip.

And which format is okay (camera/front/image or camera/front/image/compressed,

the simple image one.

Please also update your GitHub for the camera sections.

I have updated to add the removal of the CATKIN_IGNORE

Why is the front camera shown like this(please check the screenshot)? How can I get simple 2D images?

The cameras installed in the go1 are fish eye cameras and that is why they are being displayed as this. To fix it you can use opencvs filter to first crop and then remove the fisheyes distortion.

Dear @Sohail

Can you guide me a little bit more?

Should I take a chessboard and start fixing this matter?

Is there any information about the camera matrix (K) and distortion coefficients (D) for calibration?

Yes, you can find in the newly updated unitree go1 docs the info of obtaining the corrected images in here.

Dear @Sohail

I checked carefully the link which you sent me. Honestly, I was confused about how this link could help me fix the fisheye distortion. I could only find trans_rect_config.yaml, which is talking about camera distortion. Here, something is not clear to me since I am using your packages for receiveing camera data. Here I have a few questions which I would like to ask you.

  1. Should I modify anything from trans_rect_config.yaml or any other other files to access correct images?
  2. Should I modify any file from the go1_camera folder to fix this?
  3. or should I redownload the UnitreeCameraSDK transfer it to the headboard, and modify trans_rect_config.yaml or any file from the go1_camera and then run roslaunch go1_camera camera_13.launch?

Can you explain a little bit more?

Dear @Amir_Mahdi

We are not altering the camera data stream; our approach involves simply visualizing the camera input in our drivers. If you wish to modify the incoming data from the camera to alter its perspective, I recommend making adjustments within the UnitreeCameraSDK as we are also using the same SDK for our driver.

Dear @Azib

Thank you for your message. Can you guide me on what should be modified precisely to change the perspective?

I went to this directory: ~/Unitree/sdk/UnitreeCameraSdk/ for all the devices (I need camera data from front camera). I checked this link as @Sohail said. Whiting this directory ~/Unitree/sdk/UnitreeCameraSdk/, no file is called trans_rect_config.yaml (I only see a file which is called stero_camera_config.yaml). On unitree website, they mentioned that I should modify trans_rect_config.yaml as follows:

% YAML:1.0


[pls dont change] The log level
LogLevel: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 1. ]
[pls dont change] The threshold is applied to detecd point cloud
Threshold: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 190. ]
#[pls dont change] it’s a switch for a algorithm in the process of computing stereo disparity
Algorithm: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [1. ]
#UDP address for image transfer 192.168.123.IpLastSegment
IpLastSegment: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 15. ]
#This place should be changed to the actual receiving IP
#########################################
#DeviceNode
DeviceNode: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 1. ]
#To change the device number to what is needed, you can refer to the above
#########################################
#fov (perspective 60~140)
hFov: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 90. ]
#image size ([1856,800] or [928,400])
FrameSize: !!opencv-matrix
rows: 1
cols: 2
dt: d
data: [ 1856., 800. ]
#rectified frame size
RectifyFrameSize: !!opencv-matrix
rows: 1
cols: 2
dt: d
data: [ 928., 800. ]
#FrameRate, it gives the limitation for transmission rate(FPS)
FrameRate: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 3e+01 ]
#0 ori img - right 1 ori img - stereo 2 rect img - right 3 rect img - stereo -1 不传图
Transmode: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 2. ]
#The transmode here, 0 and 1 cannot be used, you need to set 2 or 3
#########################################
#Transmission rate(FPS) in the UDP transmitting process. <= FrameRate
Transrate: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 3e+01 ]
[pls dont change ]It’s a switch in distortion process of fisheye camera. 1 represents “Longitude and latitude expansion of fisheye camera”; 2 represnets “Perspective distortion correction”.
Depthmode: !!opencv-matrix
rows: 1
cols: 1
dt: d
data: [ 1. ]
empty reserved IO
Reserved: !!opencv-matrix
rows: 3
cols: 3
dt: d

If you can tell me what I should modify to fix this issue, that would be great!

Dear @Azib & @Sohail

Can you help me with the previous post?

Thank you in advance for your help :slight_smile:

Best regards,
Amir

Dear @Amir_Mahdi

Regrettably, our involvement in the camera stream is limited to its publication on RViz, and we haven’t undertaken any additional development beyond that. Therefore, we are unable to provide assistance in making specific changes, as such modifications would constitute a comprehensive development project.

I recommend reviewing the manuals, as they offer insights into the camera’s functionality. Familiarizing yourself with the manuals will not only enhance your understanding of the camera but also facilitate any adjustments you wish to make to the camera stream.

Dear @Azib

Thank you for your reply. Okay, I’ll try to figure out. But can I ask you for any hint the you technical team can tell me to speed up this process? (What should I look for to anything else that help me to understand a little bit more)

Best regards,
Amir

Dear @Amir_Mahdi

We would love to help you out but unfortunately even to give you some insightful hits we would have to go through the whole driver that itself makes a new project for us.
I hope that you understand our circumstances.