PCL-docker image

Links

GitHub repo
Docker Hub image repo

Overview

This image is based on Ubuntu 18.04 and has the following packages installed:

• CUDA-10.2.89
• nvidia-docker
• CMake-3.10.2
• VTK-8.2.0
• PCL-1.11.0
• OpenCV-4.4.0
• Eigen
• Flann
• Boost
• gdb
• openssh
• Sublime Text
• zsh

Default user is pcl and password is pcl too.
It starts an ssh server and exposes it in port 2222 of the host.

The idea is that the docker image will contain all the required software and libraries to develop on PCL and then we will connect from an IDE (CLion in my case) running in the host that will do "remote" compiling. It even allows to launch GUI applications from the container using xhost

You will need to have the nvidia GPU drivers and nvidia CUDA drivers in your host to take full advantage of this image

The image is based on tkkhuu's one with some additions to make CLion remote development work

Clion configuration

Follow the steps here.
Note: CMake will be installed on /usr/local/bin/cmake instead of on the default location

Docker image

Clone the repo and go to its folder

$ git clone https://github.com/DLopezMadrid/pcl-docker
$ cd pcl-docker

After that, you can either choose to build the image by yourself or pull it from docker hub

Pull from docker-hub (Option A)

$ docker pull dlopezmadrid/pcl-docker:latest

Build it for yourself (Option B)

Build the image with

$ ./build_image.sh

Using the image

Once the image is ready, you can create a container called pcl-docker with

$ ./setup_container.sh  

Once the container is created, there is no need to create it again unless it is deleted

Start the container with

$ ./start_container.sh

Build a PCL Project

Start the container with

$ ./start_container.sh

Then run these commands within the container

$ cd docker_dir/example_project/cloud_viewer
$ mkdir build && cd build
$ cmake ..
$ make

You can run the project then with

$ ./cloud_viewer

You may need to zoom out to see the example point clouds

Other stuff

Run a command from the container and detach it

• Run the command with &
• Press ctrl+z
• Run bg
• Run disown

Check that the container is seeing the nvidia GPU

$ nvidia-smi

Check that the host has nvidia & CUDA drivers installed

$ nvidia-smi
$ nvcc --version

Solve the bad x server connection issue

$DISPLAY is not set properly in the container In the host, run:

$ echo $DISPLAY

You should see something like :0 or :1. Then inside the docker run with the right output from before:

$ export DISPLAY=:0