[ORB SLAM2/LDSO] Docker Images

This post is originated from HeejoonLee, rewritten by YoungJ-Baek

Docker Images

1. Supported Frameworks(Datasets)

• ORB-SLAM2(Kitti)
• LDSO(Kitti, EuRoC)

2. Prerequisites

2.1. Docker Engine

How to install

2.2. Dataset

• Kitti
• EuRoC

2.3. Supported OS

x86_64 System:

• Windows 10(WSL2)
• Ubuntu

💡 This system is not compatible with ARM64 system (observed with Apple M1 Pro)

2.4. Docker Image Download

$ docker pull heejoon1130/slam:0.2

💡 Image Size: 3.84GB

2.5. Dataset Directory Setup

The datasets directory should be set up in the host system as in the following structure:

datasets
└── kitti
    └── odometry_gray
        └── sequences
            ├── 00
            ├── 01
            ├── 02
            ├── 03
            ...
└── euroc
    ├── MH_01_easy
    ├── MH_02_easy
    ├── MH_03_medium
    ├── MH_04_difficult
    ├── MH_05_difficult
    ...

3. Basic Usage (Run Docker Container)

3.1. Windows(WSL2)

$ docker run -v [HOST_DATASET_PATH]:/home/slam/datasets \
-e DISPLAY=$DISPLAY \
-it heejoon1130/slam:0.2

SLAM frameworks in the Docker image use the Pangolin viewer to display the localization and mapping process using GUI. An X server must be set up on Windows(Xming, …) and the DISPLAY variable correctly set to the host display in order to run GUI applications in WSL2.

3.2. Ubuntu

$ xhost +
$ docker run -v [HOST_DATASET_PATH]:/home/slam/datasets \
-v /tmp/.X11-unix:/tmp/.X11-unix \
-e DISPLAY=$DISPLAY \
-it heejoon1130/slam:0.2

💡 xhost + means that X server allows the connection from all clients. If you worry about the security, you can block the connection again via typing xhost - after shutting down the Docker container.

💡 HOST_DATASET_PATH is the path to the datasets directory in the host system that is mounted in /home/slam/datasets inside the Docker container

For example, if we assume that the Kitti dataset is stored in the host as follows:

~/shared/datasets/kitti/odometry_gray/sequences

Then, HOST_DATASET_PATH should be set in this way, ~/shared/datasets

3.3. Run SLAM

At the first run, options for selecting framework and dataset will be printed. Then, you can choose appropriate options and run the SLAM.

Run example:

$ docker run -v ~/shared/datasets:/home/slam/datasets -e DISPLAY=$DISPLAY -it heejoon1130/slam:0.2
[V-SLAMMERS SLAM Docker Image]

Select the SLAM framework to use
Enter 'q' to quit
(1) ORB-SLAM 2
(2) LDSO
Enter a number: 2
---------------------------
Running LDSO
Select a dataset:
(1) Kitti
(2) EuRoC
Enter a number: 1
Select a sequence(Default:00):
['00', '01', '02', '03', '04', '05', '06', '07', '08', '09', '10', '11', '12', '13', '14', '15', '16', '17', '18', '19', '20', '21']
Sequence: 00

To terminate, input ctrl + c → q → Enter

4. Advanced Usage

4.1. Run Docker Container with bash

$ docker run -v [HOST_DATASET_PATH]:/home/slam/datasets \
-e DISPLAY=$DISPLAY \
-it heejoon1130/slam:0.2 bash

The output will be like as followed.

root@23kflkjasd8f:/home/slam#

Then, you may modify the source code directly via terminal and test the changes you have made.