[Unitree Go1][Camera] Accessing the camera

Sensors
1

Hello @Sohail ,

I am trying to access the camera of go1.
I proceeded as per the document

  • killing the default (i have doubt if it is killed properly)
  • Running the camera launch files

I am able to get the images and visualize them in rviz.

The problem is :

  • A single image consist of both left and right camera output
  • The enclosure of camera is also captured
  • The images are in fish eye format
  • The rate of publishing is slow. around 3Hz

Go1_NativeFrontCamera_Output
Go1_NativeFrontCamera_Output1852×807 119 KB

Kill output

>13unitree@unitree-desktop:~$ ./Unitree/autostart/camerarosnode/cameraRosNode/kill.sh

Usage:
 kill [options] <pid> [...]

Options:
 <pid> [...]            send signal to every <pid> listed
 -<signal>, -s, --signal <signal>
                        specify the <signal> to be sent
 -l, --list=[<signal>]  list all signal names, or convert one to a name
 -L, --table            list all signal names in a nice table

 -h, --help     display this help and exit
 -V, --version  output version information and exit

For more details see kill(1).
>13unitree@unitree-desktop:~$

Do you know a way to get proper rectangular single combined images?
How to get depth information?
Is there any other packages available for the same.

And do you have any suggestions for object tracking packages for Go1? (like yolo)

Thank you,
Vishnu

2

Dear @Vishnu,

Point 1/2/3 I believe there is no out of the box solution provided by Unitree for auto corrected images. Most likely you would have to perform image correction and cropping techniques from opencv. You can always try the Unitreecamera SDK and I think they have a software that performs the correction but I cannot recall and the process was not straight forward.

Point 4 To accelerate the speed of the topics, depends upon the connection type you have LAN or WiFi. The 3hz is slow and seems to be over WiFi. For this hardware connections matter to accelerate it, you can also directly edit the provided packages to up the publishing speed but be careful to not flood the network as it may affect the other ROS control commands.

For Depth information, your best bet would be using stereo to depth packages (just google it and there are plenty open source packages available including ones directly from opencv) and/or simply use intel realsense and/or zed2i to get that information.

For object tracking I would recommend using yolo due to its software stability, ease of use, and small processing footprint. Other than there is darknet which was quite good for ROS several years back.

1 Like