Docker Images for ZED Camera

The repository contains a GitHub Actions workflow for auto-deployment of built Docker images to husarion/zed-desktop, husarion/zed-desktop-cpu and husarion/zed-jetson Docker Hub repositories. This process is based on zed-ros2-wrapper repository.

Available Docker Images

ROS1:

• husarion/zed-desktop:noetic for desktop platform with CUDA (tested on platform with 11.7).
• husarion/zed-desktop-cpu:noetic a simple demo version on the CPU, allowing you to check if the camera and IMU are working properly. Image dedicated to platforms with amd64 and arm64 architecture.
• husarion/zed-jetson:noetic for Jetson platform currently support - Jetson Xavier, Orin AGX/NX/Nano, CUDA 11.4 (tested on Xavier AGX).
• husarion/zed-jetson:melodic for Jetson platform currently not supported - Jetson Nano, TX2/TX2 NX, CUDA 10.2 (tested on Nano).

ROS2:

• husarion/zed-desktop:humble for desktop platform with CUDA (tested on platform with 11.7).
• husarion/zed-desktop-cpu:humble a simple demo version on the CPU, allowing you to check if the camera and IMU are working properly. Image dedicated to platforms with amd64 and arm64 architecture.
• husarion/zed-jetson:foxy for Jetson platform currently support - Jetson Xavier, Orin AGX/NX/Nano, CUDA 11.4 (tested on Xavier AGX).

Prepare Environment

1. Plugin the Device

For best performance please use USB 3.0 port, depend of the camera model. Then use lsusb command to check if the device is visible.

Important

To use the ZED X camera, you'll need to install the ZED X Driver additionally. Follow the official tutorial Setting up ZED X on Orin Nano / NX Developer Kits provided by Stereolabs for detailed instructions.

Demo GPU

1. Clone the repository

git clone https://github.com/husarion/zed-docker.git
cd zed-docker/demo

2. Select Appropriate Docker Image

export ZED_IMAGE=<zed_image>

Replace <zed_image> with appropriate Docker image from below table.

Docker Image
husarion/zed-jetson:foxy
husarion/zed-desktop:humble

3. Select the appropriate camera model

export CAMERA_MODEL=<camera_model>

Replace <camera_model> with appropriate camera model from below table.

Product Name	Camera Model
ZED	zed
ZED Mini	zedm
ZED 2	zed2
ZED 2i	zed2i
ZED X	zedx
ZED X Mini	zedxm

4. Activate the Device

docker compose up zed

Note

You can run the visualization on any device, provided that it is connected to the computer to which the sensor is connected.

5. Launch Visualization

xhost local:root
docker compose up rviz

Note

First run of ROS 2 Docker images downloads configuration files and optimize camera. This may take several minutes.

6. Configure RViz.

Add following topic to RViz:

• RGB image topic: /<camera_model>/zed_node/rgb/image_rect_color
• Point cloud topic: /<camera_model>/zed_node/point_cloud/cloud_registered

Demo CPU

1. Clone the repository

git clone https://github.com/husarion/zed-docker.git
cd zed-docker/demo

2. Activate the Device

docker compose -f dompose-cpu.yaml up

3. Launch Visualization

xhost local:root
docker compose up rviz

Note

First run of ROS 2 Docker images downloads configuration files and optimize camera. This may take several minutes.

Additional details

Docker Compose volume - /tmp/argus_socket:/tmp/argus_socket is only necessary for ZED X camera.

Known issue

Docker Buildkit doesn't support build image with Nvidia runtime, so there is dedicated action for pushing locally building image from Jetson.