ig_lio_workspace

This Docker is a standard development environment for iG-LIO, which helps you quickly experience iG-LIO. The Docker is visualized via VNC, noVNC, and Xserver, allowing users to see the trajectory and mapping in real time.

Supported scheme	Viusal manner
Ubuntu (any version) with GPU (recommended)	Xserver
Ubuntu without GPU	VNC or noVNC
Windows with GPU	Xserver
Windows without GPU	VNC or noVNC

Since Docker utilizes the CPU to render frames in VNC, RViz may experience delays when visualizing dense point clouds. It is advisable to use sparse_map(VNC) to visualize the global map in RViz.

1. Quick Start

1.1 Install Docker

Please install Docker following the links. This Docker supports both Ubuntu and Windows.

Ubuntu: install Docker engine and nvidia-docker.
Windows (without GPU): If you plan to visualize in VNC, just install Docker desktop.
WIndows (with GPU): If you want to GPU acceleration, you should install docker engine in WSL2, and then install NVIDIA Container Toolkit on WSL2. It is recommended that follow this tutorial.

1.2 Pull Container

docker pull zijiechenrobotic/ig-lio-workspace:latest

# for Chinese mainland
docker pull registry.cn-guangzhou.aliyuncs.com/zijiechenrobotics/ig-lio-workspace:latest

1.3 Launch Container

If you are not interested in the Docker's source code, you do not need to git clone. Simply create a new workspace in the following format.

# the "lidar_dataset" folder is used to store the datasets

# download iG-LIO source code
cd workspace
git clone https://github.com/zijiechenrobotics/ig_lio.git
git clone https://github.com/Livox-SDK/livox_ros_driver.git

1.3.1 Launch Container in Ubuntu with GPU (recommend)

Enter the below contents in the docker-compose.yml

version: '3.4'
services:
  ig-lio-workspace:
    build:
      context: .
    image: zijiechenrobotic/ig-lio-workspace:latest
    # or
    # registry.cn-guangzhou.aliyuncs.com/zijiechenrobotics/ig-lio-workspace:latest
    environment:
      - DISPLAY=$DISPLAY
      - QT_X11_NO_MITSHM=1
      - NVIDIA_DRIVER_CAPABILITIES=all
      - NVIDIA_VISIBLE_DEVICES=all
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [ gpu ]
    volumes:
      # Map ig-lio source code to Docker (please change to your own path)
      - <your workspace folder path>:/root/workspace
      # Map datasets to Docker (Please change to your own path)
      - <your lidar_dataset folder path>:/root/lidar_dataset
      - /tmp/.X11-unix:/tmp/.X11-unix:rw

Launch container

# cd to ig_lio_workspace
docker compose up -d

Shut down container

# cd to ig_lio_workspace
docker compose down

Access Container

# enter below command in the local terminal (necessary)
xhost +local:root

# access container
docker exec -it <containter id> bash

# check opengl is rendered by GPU
inxi -G
# output-> OpenGL: renderer: <your GPU namae> instead of llvmpipe

1.3.2 Launch Container in Windows with GPU

Coming soon...

1.3.3 Launch Containter with VNC

version: '3.4'
services:
  ig-lio-workspace:
    build:
      context: .
    image: zijiechenrobotic/ig-lio-workspace:latest
    # or 
    # registry.cn-guangzhou.aliyuncs.com/zijiechenrobotics/ig-lio-workspace:latest
    environment:
      - VNC_PW=abc123 # user defined password
      - VNC_GEOMETRY=1280x720 # vnc resolution
      - VNC_DEPTH=24 # 16/24/32
    volumes:
      # Map ig-lio source code to Docker (please change to your own path)
      - <your workspace folder path>:/root/workspace
      # Map datasets to Docker (Please change to your own path)
      - <your lidar_dataset folder path>:/root/lidar_dataset
    ports:
      # noVNC port:
      - 46080:6080
      # standard VNC port:
      - 45901:5901

Launch container

# cd to ig_lio_workspace
docker compose up -d

Shut down container

# cd to ig_lio_workspace
docker compose down

Access via VNC

Enter the following address to access Docker via any VNC viewer ( recommend Windows uses MobaXterm and Ubuntu uses Remmina)

localhost:45901

Access via noVNC (recommend)

Any browser, enter the following address in the address bar

localhost:46080/vnc.html

1.5 Run

# build iG-LIO
cd /root/workspace/ig_lio
catkin_make

# launch iG-LIO
source devel/setup.bash
roslaunch ig_lio lio_<dataset name>.launch

Since we have mapped the datasets to the lidar_dataset, we can play the datasets inside Docker

# open another terminal
cd /root/lidar_dataset
rosbag play <dataset name>

Docker Environment

base: nvidia/opengl:1.0-glvnd-runtime-ubuntu20.04
ROS noetic
evo
gcc9 & g++9
glog
git

zijiechenrobotics/ig_lio_workspace