Carrier board mini - cameras not detected

Hi,

I have a fresh install on the mini carrier board with Jetson Orin NX 8GB, according to these instructions:

When I try to run the diagnostics, no cameras are detected. When running ZED_Explorer --all, it returns (Argus) Error Timeout: (propagating from src/rpc/socket/client/SocketClientDispatch.cpp, function openSocketConnection(), line 220)
(Argus) Error Timeout: Cannot create camera provider (in src/rpc/socket/client/SocketClientDispatch.cpp, function createCameraProvider(), line 107)
No ZED detected

Diagnostic files generated with sudo ZED_Diagnostic --dmesg and sudo ZED_Diagnostic -c below.

dmesg.log (11.3 KB)

ZED_Diagnostic_Results.json (19.3 KB)

Best,

Andreas

Hi @andrwr_rev
Have you connected the camera or modified the hardware setup after booting the device?

I have not changed the hardware setup in any way after booting the device.

Some more information though, it seems like the issue occurs after installing NVidia Deepstream according to the instructions here.

If the Deepstream installer modifies the DTB setup, then you must reinstall the ZED X Driver to fix if.

We are going to containerize our application to work around this

1 Like

Ok, so we immediately ran into issues with running the OpenCV calibration tool in the containerized version (this topic). As part of workaround attempts I installed the ZED SDK outside the container, and it immediately caused the same issues as in my original post. ZED_Diagnostic fails with the log below.

dmesg.log (11.3 KB)

Removing and reinstalling the driver and SDK does not solve the issue. That is, the Diagnostics will always fail until the carrier mini is flashed again. Please advise @Myzhar

I met internally with the team to try to understand more about this problem.
We used Deepstream internally, and we have not faced issues, so there’s something in your installation process that is breaking the GMSL2 functionalities.

If Deepstream modified the DTB or the kernel, you should have made the ZED X Camera work again by reinstalling the driver.
The fact that this does not work makes us believe that the Argus libraries, at some point, have been modified.

Can you please share the procedure that you are following to configure the Jetson to install and configure Deepstream (or anything else that is not ZED-related)?

Sure, here is the procedure:

  • Initial hardware setup: Carrier Mini W/Orin NX 8GB. 2x ZED X One connected.
  • Flash the card according to the instructions for Zed box mini.
  • sudo apt update , sudo apt upgrade -y, sudo apt-dist upgrade
  • Reboot
  • sudo apt install nvidia-jetpack
  • Install driver 1.4.0
  • Install ZED SDK 5.2
  • Install the following (probably some duplicates, they are installed over multiple scripts)
    • sudo

    • apt-utils

      build-essential

      ccache

      cmake

      curl

      doxygen

      git

      gpg-agent

      software-properties-common

      cargo

      net-tools

      wget

      lsb-release

      less

    • python3-pip

      python3-dev

    • clang-format

      cmake

      gdb

      libprotobuf-dev

      nano

      protobuf-compiler

      tmux

      valgrind

      vim

  • Install zenoh according to These instructions
  • Install deepstream 7.1 for Jetson (we use the .deb): Installation — DeepStream documentation
    • This also modifies glib and installs kafka
    • Also installs the below
    • libssl3 \
      libssl-dev \
      libgstreamer1.0-0 \
      gstreamer1.0-tools \
      gstreamer1.0-plugins-good \
      gstreamer1.0-plugins-bad \
      gstreamer1.0-plugins-ugly \
      gstreamer1.0-libav \
      libgstreamer-plugins-base1.0-dev \
      libgstrtspserver-1.0-0 \
      libjansson4 \
      libyaml-cpp-dev \
      libmosquitto1

What method of the 4 have you used to install DeepStream?

Why have you installed Zenoh? (This is a curiosity of mine…)

Method 3, the .deb install.

As far as Zenoh goes, it is our choice of protocol for pub/sub comms between the Jetson device and a separate CPU (No, we do not use ROS)

2 Likes

This is the same method we tested with no issues.

Interesting choice :+1:

I have to prioritize an upcoming tech demo and will not follow this up immediately, insofar as our current containerized setup seems to be working at the moment.

1 Like