仓库链接
UZ-SLAMLab/ORB_SLAM3

系统

我前面弄了Apollo所以想再整个新系统来弄SLAM，也可以直接在原系统中进行。

  1. 在wsl2-distro-manager的release中下载最新版的压缩包。
  2. 打开就可以复制已有的WSL。

依赖

Pangolin

Pangolin-Github

1
2
3
4
5
6
7
8
9
10
git clone https://github.com/stevenlovegrove/Pangolin
cd Pangolin/scripts/
rm -rf vcpkg/
git clone https://github.com/microsoft/vcpkg
cd ..
./scripts/install_prerequisites.sh --dry-run recommended
mkdir build && cd build
cmake ..
cmake --build .
sudo make install

OpenCV

opencv.org

  1. 安装依赖 
    1
    2
    3
    sudo apt install build-essential libgtk2.0-dev libavcodec-dev libavformat-dev libjpeg-dev libswscale-dev libtiff5-dev
    sudo apt install libgtk2.0-dev
    sudo apt install pkg-config
  2. 安装OpenCV 
    1
    2
    3
    4
    5
    6
    7
    8
    sudo apt update && sudo apt install g++ unzip wget cmake
    wget -O opencv.zip https://github.com/opencv/opencv/archive/refs/tags/4.4.0.zip
    unzip opencv.zip
    cd opencv-4.4.0/
    mkdir build && cd build
    cmake ..
    cmake --build .
    sudo make install

    安装好依赖的话在5.3中应该就不会报错

Eigen3

eigen.tuxfamily.org

1
2
3
4
5
6
wget https://gitlab.com/libeigen/eigen/-/archive/3.1.1/eigen-3.1.1.zip
unzip eigen-3.1.1.zip && cd eigen-3.1.1
mkdir build && cd build
cmake ..
sudo make install
sudo cp -r /usr/local/include/eigen3/Eigen /usr/local/include

不知道为什么这个不用cmake --build .

安装

Clone the repository:

1
git clone https://github.com/UZ-SLAMLab/ORB_SLAM3.git ORB_SLAM3
1
2
3
cd ORB_SLAM3
chmod +x build.sh
./build.sh

报错1

1
2
/home/trrrrw/orb-slam3/ORB_SLAM3/Thirdparty/DBoW2/DBoW2/BowVector.h:17:10: fatal error: boost/serialization/serialization.hpp: No such file or directory
 #include <boost/serialization/serialization.hpp>
1
sudo apt install libboost-dev

报错2

1
2
/home/trrrrw/orb-slam3/ORB_SLAM3/src/System.cc:26:10: fatal error: openssl/md5.h: No such file or directory
 #include <openssl/md5.h>
1
sudo apt install libssl-dev

报错3

1
/usr/bin/ld: cannot find -lboost_serialization

安装boost

1
sudo apt install libboost-all-dev

测试数据集

  1. 下载数据集
    https://projects.asl.ethz.ch/datasets/doku.php?id=kmavvisualinertialdatasets#downloads
    下载ASL格式
    放在最外面新建文件夹Examples

    能找到就行

  2. 解压出mav0文件夹
  3. 在ORB_SLAM3下新建文件夹dataset/MH01，将mav0文件夹放在这 
    1
    2
    3
    4
    mkdir dataset && cd dataset
    mkdir MH01 && cd MH01
    # 把mav0文件夹复制到MH01里就行
    cp -r ../../../Examples/mav0/ ./
  4. 运行 
    1
    2
    cd ../../Examples
    ./Monocular/mono_euroc ../Vocabulary/ORBvoc.txt ./Monocular/EuRoC.yaml ../dataset/MH01 ./Monocular/EuRoC_TimeStamps/MH01.txt

    这样应该是没有窗口的
    在 Examples/Monocular/mono_euroc.cc 第 83 行附近，将函数的最后一个参数更改为 true ORB_SLAM3::System SLAM(argv[1],argv[2],ORB_SLAM3::System::MONOCULAR, true); 默认情况下为 false，改为true
    用WSL这样还是没有窗口🤷‍♂️

WSL2

  1. 安装xfce4 
    1
    sudo apt install xfce4
  2. 重新编译会报错（如果2.2.1没安装依赖的话） 
    1
    what():  OpenCV(4.4.0) /home/trrrrw/orb-slam3/opencv-4.4.0/modules/highgui/src/window.cpp:634: error: (-2:Unspecified error) The function is not implemented. Rebuild the library with Windows, GTK+ 2.x or Cocoa support. If you are on Ubuntu or Debian, install libgtk2.0-dev and pkg-config, then re-run cmake or configure script in function 'cvNamedWindow'
    1. 删除OpenCV后安装libgtk2.0-dev and pkg-config后重新编译安装 
      1
      2
      cd opencv-4.4.0/build
      sudo make uninstall
    2. 安装依赖 
      1
      2
      3
      sudo apt install build-essential libgtk2.0-dev libavcodec-dev libavformat-dev libjpeg-dev libswscale-dev libtiff5-dev
      sudo apt install libgtk2.0-dev
      sudo apt install pkg-config
    3. 编译安装 
      1
      2
      3
      4
      cd build
      cmake ..
      cmake --build .
      sudo make install
  3. 再执行4.4
    成功👍
    成功运行