PLSVO（单目）

Posted 2020-10-11

代码：https://github.com/rubengooj/pl-svo

安装SVO，with ROS：http://blog.sina.com.cn/s/blog_7b83134b0102wfu4.html

安装SVO，without ROS，参考下面：

1.Prerequisites and dependencies

(1)SVO

See the SVO Wiki for more instructions: https://github.com/uzh-rpg/rpg_svo/wiki 

选择：No ROS (plain CMake)

[1]Boost，Eigen 3，OpenCV，Sophus,g2o都已经安装过.

[2]安装Fast - Corner Detector

cd workspace
git clone https://github.com/uzh-rpg/fast.git
cd fast
mkdir build
cd build
cmake ..      //编译（cmake---编译工具）
make
sudo make install //安装，可以在/usr/local/include/fast/；/usr/local/lib/libfast.so找到

[3]安装vikit_common - Some useful tools that we need

cd workspace
git clone https://github.com/uzh-rpg/rpg_vikit.git
cd rpg_vikit/vikit_common
mkdir build
cd build
cmake ..
make

(2)MRPT---In case of using the provided representation.

sudo apt-get install libmrpt-dev

(3)Line descriptor (in 3rdparty folder)

We have modified the line_descriptor module from the OpenCV/contrib library (both BSD) which is included in the 3rdparty folder.

2.Configuration and generation---配置和生成

运行build.sh文件配置和生成the line_descriptor模块，然后会配置和生成PL-SVO库---libplsvo.so，在lib文件夹中；生成了可执行文件bin/run_pipeline。

此过程中出现了错误：

/home/wj/Downloads/pl-svo/include/plsvo/sceneRepresentation.h:119:5: error: ‘CVectorDouble’ does not name a type
     CVectorDouble   v_aux, v_aux_, v_aux1, v_aux1_, v_auxgt, gt_aux_, v_auxgt_;

修改：找到这个头文件sceneRepresentation.h，将 CVectorDouble换成 mrpt::vector_double 即可。

3. Dataset format and usage

The run_pipeline basic usage is:

./run_pipeline  <dataset_path>  

在< dataset_path >指序列文件夹相对于环境变量${DATASETS_DIR}，必须预先设定。序列文件夹必须包含数据集配置文件命名following the examples in pl-svo/config,其中images_subfolder refers to the image subfolder.（图像子文件夹）