ORB-SLAM3 Docker Setup (CUDA + X11(GUI) Support + Ubuntu 22.04)

This guide will walk you through setting up ORB-SLAM3 in a Docker container, running it with a EuRoC dataset, and testing it with different configurations like Monocular, Monocular-Inertial, and Stereo.

After setting up the Docker container, you can run the ORB-SLAM3 examples with GPU acceleration and X11 forwarding for graphical output that will look like this:

You will also have ability to attach to a running container and compile code using ORB SLAM 3 library. This setup is more for development purposes and testing different configurations of ORB-SLAM3.

Prerequisites

• Docker installed
• NVIDIA GPU and drivers with CUDA support for GPU acceleration
• X11 server running for displaying graphical applications (e.g., VcXsrv for Windows). Material for setting up X11 server on Windows is also included below.

Getting Started

Step 1: Running the Docker Container

First you need to build the Docker image. You can do this by running the following command in the root directory of this repository:

docker build -t orb_slam3:latest .

To run the ORB-SLAM3 Docker container with GPU support and X11 forwarding for graphical output:

docker run -it --gpus all --env="DISPLAY" \
  --env="QT_X11_NO_MITSHM=1" \
  --volume="/tmp/.X11-unix:/tmp/.X11-unix:rw" \
  --name orb_slam3_container \
  orb_slam3:latest

Step 2: Dataset Setup

After setting it up container and starting it, you need to attach to it (quick guide how to do it is below). Navigate to the ORB-SLAM3 directory and download the EuRoC MAV dataset. This dataset is used for testing ORB-SLAM3 with different configurations.

cd /opt/orb_slam3

# Create directory for dataset
mkdir -p Datasets/EuRoc
cd Datasets/EuRoc/

# Download the dataset
wget -c http://robotics.ethz.ch/~asl-datasets/ijrr_euroc_mav_dataset/machine_hall/MH_01_easy/MH_01_easy.zip

# Create a directory for MH_01_easy dataset and extract it
mkdir MH01
unzip MH_01_easy.zip -d MH01/

Step 3: Installing Additional Packages

To run graphical applications inside the container (e.g., xclock for testing display):

apt-get update && apt-get install -y x11-apps

Test the X11 setup by running:

xclock

You should see the clock displayed on your X11 server.

Step 4: Running ORB-SLAM3 Examples

To run different ORB-SLAM3 configurations, use the following commands inside ORB-SLAM3 directory (which is located inside /opt inside the container):

Stereo Example

./Examples/Stereo/stereo_euroc ./Vocabulary/ORBvoc.txt \
  ./Examples/Stereo/EuRoC.yaml \
  ../Datasets/EuRoc/MH01 \
  ./Examples/Stereo/EuRoC_TimeStamps/MH01.txt dataset-MH01_stereo

Step 5: Setting Up X11 Server on Windows

If you're running this on Windows, you can set up an X11 server using VcXsrv. Download and install it from the following link:

• VcXsrv SourceForge

Ensure that DISPLAY is correctly set in your Docker run command to match your X11 server configuration. It is also recommended to disable access control on the X11 server to avoid permission issues:

If using win app you just need to check the box "Disable access control" in the VcXsrv configuration.

Alternatively: If using linux based environment, you can use the following commands to grant access to the X11 server:

export DISPLAY=host.docker.internal:0
xhost +

If you want to test the X11 setup, you can run the following command inside the container:

xclock

This should display a clock on your X11 server. If you see the clock, the X11 setup is working correctly. If not, check your X11 server configuration and Docker run command.

Attaching to a Running Container

If you want to attach to a running container, you can use the following command:

docker exec -it orb_slam3_container bash

This will open a new shell inside the running container.

Alternatively: You could use vscode to attach to the container, you just need to install the extension "Dev - Containers" and open the project folder in the container.

If you want to attach to a running container, you can use the following command:

ctrl + shift + p
Remote-Containers: Attach to Running Container

Select /opt/ as the workspace folder. Later if you want to change the workspace folder you need to change configuration with the following steps:

ctrl + shift + p
Remote-Containers: Open Container Configuration File