OpenVSLAM的学习

一、OpenVSLAM主要工作

论文：Sumikura S, Shibuya M, Sakurada K. OpenVSLAM: A Versatile Visual SLAM Framework[C]//Proceedings of the 27th ACM International Conference on Multimedia. 2019: 2292-2295.
代码：https://github.com/xdspacelab/openvslam ；文档

创新：

• 支持多种相机的模型；
• 实现地图保存和加载，可以根据新的帧定位到保存的地图中的位置；
• 首个实现广角相机的开源SLAM框架；
• 相比于ORB-SLAM2，暴力匹配是可选择性的，所以匹配更加准确。

与其他框架的对比如下图所示：

部分算法原理：

• 前端使用非直接法，提取ORB特征点
• 具有三个线程：跟踪、建图和全局优化

二、OpenVSLAM的编译

git clone https://github.com/xdspacelab/openvslam

环境需求：

• Eigen : version 3.3.0 or later.
• g2o : Please use the latest release. Tested on commit ID 9b41a4e.
  SuiteSparse : Required by g2o.
• DBoW2 : Please use the custom version of DBoW2 released in https://github.com/shinsumicco/DBoW2.
• yaml-cpp : version 0.6.0 or later.
• OpenCV : version 3.3.1 or later.

安装依赖：

sudo apt-get update
sudo apt-get install build-essential pkg-config cmake git wget curl unzip
sudo apt-get install libatlas-base-dev libsuitesparse-dev
sudo apt-get install libgtk-3-dev
sudo apt-get install ffmpeg
sudo apt-get install libavcodec-dev libavformat-dev libavutil-dev libswscale-dev libavresample-dev
sudo apt-get  install libgoogle-glog-dev libgflags-dev
sudo apt-get install libopenblas-dev
sudo apt-get install --no-install-recommends libboost1.65-all-dev 
sudo apt-get install libx11-dev 
sudo apt-get install libgl1-mesa-dev 
sudo apt-get install freeglut3-dev 
sudo apt-get install doxygen 
wget https://nchc.dl.sourceforge.net/project/glew/glew/2.1.0/glew-2.1.0.tgz --no-check-certificate 
tar -xzvf glew-2.1.0.tgz  && cd glew-2.1.0/ 
make -j2
sudo make install 
sudo ln -s /usr/lib64/libGLEW.so.2.1 /usr/lib/libGLEW.so.2.1
sudo ldconfig -v

build工程：

cd 你的目录/openvslam
mkdir build && cd build
cmake \
    -DBUILD_WITH_MARCH_NATIVE=ON \
    -DUSE_PANGOLIN_VIEWER=ON \
    -DUSE_SOCKET_PUBLISHER=OFF \
    -DUSE_STACK_TRACE_LOGGER=ON \
    -DBOW_FRAMEWORK=DBoW2 \
    -DBUILD_TESTS=ON \
    ..
make -j4

如果cmake找不到依赖：

设置绝对路径，例如 当 CMAKE_INSTALL_PREFIX是 /usr/local:

Eigen3_DIR=/usr/local/share/eigen3/cmake

OpenCV_DIR=/usr/local/share/OpenCV

DBoW2_DIR=/usr/local/lib/cmake/DBoW2

g2o_DIR=/usr/local/lib/cmake/g2o

Pangolin_DIR=/usr/local/lib/cmake/Pangolin (if installed)
sioclient_DIR=/usr/local/lib/cmake/sioclient (if installed)

编译OpenVSLAM时遇到如下问题：

/usr/bin/ld: /usr/local/lib/libyaml-cpp.a(scanscalar.cpp.o): relocation R_X86_64_PC32 against symbol `ZNK4YAML5RegEx14MatchUncheckedINS_16StreamCharSourceEEEiRKT’ can not be used when making a shared object; recompile with -fPIC
/usr/bin/ld: 最后的链接失败: 错误的值
collect2: error: ld returned 1 exit status
src/openvslam/CMakeFiles/openvslam.dir/build.make:1926: recipe for target ‘lib/libopenvslam.so’ failed
make[2]: *** [lib/libopenvslam.so] Error 1
CMakeFiles/Makefile2:138: recipe for target ‘src/openvslam/CMakeFiles/openvslam.dir/all’ failed
make[1]: *** [src/openvslam/CMakeFiles/openvslam.dir/all] Error 2
Makefile:129: recipe for target ‘all’ failed
make: *** [all] Error 2

因为：yaml-cpp defaults to building a static library, but you may build a shared library by specifying -DYAML_BUILD_SHARED_LIBS=ON.

所以在yaml-cpp的cmake选项上加上
-DYAML_BUILD_SHARED_LIBS=ON

重新编译yaml-cpp：

git clone https://github.com/jbeder/yaml-cpp
cd yaml-cpp
mkdir build && cd build
cmake \
    -DYAML_BUILD_SHARED_LIBS=ON
    -DCMAKE_BU ILD_TYPE=Release \
    -DCMAKE_INSTALL_PREFIX=/usr/local \
    ..
make -j2
sudo make install 
sudo ldconfig -v

最后重新编译openvslam即可

三、测试

TUM深度数据集
在build下运行

./run_tum_rgbd_slam \ 
 -v /自己的路径/openvslam/voc/orb_vocab/orb_vocab.dbow2 \
 -d /自己的路径/rgbd_dataset_freiburg1_360 \
 -c ../example/tum_rgbd/TUM_RGBD_rgbd_1.yaml \
 -a 自己的路径/associated.txt

Euroc单目数据集测试
同样在build下
#monocular SLAM with any EuRoC sequence

./run_euroc_slam \
-v /path/to/orb_vocab/orb_vocab.dbow2 \
-d /path/to/EuRoC/MAV/mav0/ \
-c ../example/euroc/EuRoC_mono.yaml