硬件：XbotU-bj008
系统：Ubuntu 16.04.1
ROS版本：Kinect 1.12.14
# 打开 https://github.com/johnchars/ORB_SLAM
# 下载install.bash
./install.bash-
Boost
sudo apt-get install libboost-all-dev
Boost 库用于同时启动不同的线程，ORB-SLAM分为三个线程Tracking, Local Mapping and Loop Closing
-
ROS
参考官方wiki
-
OpenCV
cd && mkdir Repo && cd Repo git clone https://github.com/opencv/opencv.git cd opencv git branch -a git checkout origin/2.4 git branch -m origin/2.4 opencv-2.4 git branch mkdir build && cd build cmake .. && make -j4 sudo make install
这里也可以使用官网下载，注意选择2.4版本; make -j*中数字可以自己决定
-
g2o
sudo apt-get install libeigen3-dev
安装g2o图优化库需要安装eigen库，使用了一个修改过的g2o库，在Thirdparty/下有这个库，不需要下载
-
DBow2
同样的第三库，在Thirdparty/下有这个库，用于回环检测。
-
创建一个安装位置
cd && mkdir -p slam_ws/src && cd slam_ws/src catkin_init_workspace cd .. && catkin_make echo "source ~/slam_ws/devel/setup.bash" >> ~/.bashrc git clone https://github.com/raulmur/ORB_slam_ws.git ORB_slam_ws
-
[TODO] 修改环境变量 ROS_PAKCAGE_PATH，在~/.bashrc末尾增加
export ROS_PACKAGE_PATH=${ROS_PACKAGE_PATH}:/home/xbot/slam_ws/src
-
编译g2o
cd ~/slam_ws/src/ORB_SLAM/Thirdparty/g2o/ mkdir build && cd build cmake .. -DCMAKE_BUILD_TYPE=Release make -j4
-
编译DBoW2
cd ../../DBoW2 mkdir build && cd build cmake .. -DCMAKE_BUILD_TYPE=Release make -j4
-
编译ORB_SLAM
删除manifest.xml 中的
在ORBextractor.cc中添加头文件，路径是/home/xbot/slam_ws/src/ORB_SLAM/src
#include <opencv2/features2d/features2d.hpp>
#include <opencv2/imgproc/imgproc.hpp>
在/home/xbot/slam_ws/Thirdparty/g2o/g2o/solvers中的linear_solver_eigen.h中修改
56: typedef Eigen::PermutationMatrix<Eigen::Dynamic, Eigen::Dynamic, SparseMatrix::Index> PermutationMatrix; 修改为 56: typedef Eigen::PermutationMatrix<Eigen::Dynamic, Eigen::Dynamic> PermutationMatrix;
在/home/xbot/slam_ws/src/ORB_slam_ws下修改CMakeLists.txt文件，增加
find_package(Boost COMPONENTS system) include_directories( ${Boost_INCLUDE_DIRS} // adding this line ) target_link_libraries(${PROJECT_NAME} ${Boost_LIBRARIES} )
开始编译...
cd ~/slam_ws/src/ORB_SLAM/ mkdir build && cd build cmake .. -DROS_BUILD_TYPE=Release make -j4 编译成功显示 Build type: RelWithDebInfo -- Boost version: 1.58.0 -- Found the following Boost libraries: -- system -- Configuring done -- Generating done -- Build files have been written to: /home/xbot/slam_ws/src/ORB_SLAM/build [ 0%] Built target rospack_genmsg_libexe [ 0%] Built target rosbuild_precompile [ 5%] Linking CXX executable ../bin/ORB_SLAM [100%] Built target ORB_SLAM
-
可能的编译错误解决
- 'FAST' was not declared in this scope FAST(cellImage, cellKeyPoints [i] [j], fastTh, true);
解决方法GitHub #44
在src/ORBextractor.cc中增加
#include <opencv2/opencv.hpp>
增加在#include <opencv2/core/core.hpp>前
- DSO missing from command line
解决方Github#552
在CMakeLists.txt中增加Boost头文件
include_directories(
${Boost_INCLUDE_DIRS} #增加这一行
)
-
cmake Configuring incomplete, errors occurred!
解决方法，检查ROS_PACKAGE_PATH路径，在.bashrc文件下
-
其他问题可以在ORB_SLAM下检索关键词
a. 解压Data下的词库和设置文件
cd /home/xbot/slam_ws/src/ORB_SLAM/Data
xbot@nuc:~/slam_ws/src/ORB_SLAM/Data$ tar xzvf ORBvoc.txt.tar.gz
ls
增加 ORBvoc.txt文件b.逐个文件启动
Ctrl+Alt+t
roscore
Ctrl+Alt+t
cd /home/xbot/slam_ws/src/ORB_SLAM
rosrun ORB_SLAM ORB_SLAM Data/ORBvoc.txt Data/Setting.yaml[todo] adding picture orb_node_success
Ctrl+Alt+t
rosrun image_view image_view image:=/ORB_SLAM/Frame _autosize:=true[todo] adding picture
cd /home/xbot/slam_ws/src/ORB_SLAM
rosrun rviz rviz -d Data/rviz.rviz[todo] adding picture orb_rviz_d
c.使用launch文件启动
cd /home/xbot/slam_ws/src/ORB_SLAM
roslaunch ExampleGroovyOrNewer.launchd. 使用Example.bag 检测安装是否成功
-
下载http://webdiis.unizar.es/~raulmur/orbslam/downloads/Example.bag.tar.gz文件
备用链接 https://drive.google.com/file/d/0B8Qa2__-sGYgRmozQ21oRHhUZWM/view?usp=sharing
-
切换到bag文件路径，这里在/home/xbot/下建立了Dataset/文件夹，并将文件放到这里
cd /home/xbot/Dataset/ tar xzvf Example.bag.tar.gz rosbag play --pause Example.bag #按空格开始
[todo] 效果图
特别感谢这篇教程，解决了一周的难受
这里需要注意的是ORB_slam_ws只接收来自名为/camera/image_raw的topic信息，如果使用单目摄像头如logitech c270i或者Realsense D415的color camera，需要修改topic名称。
首先标定摄像头的内参矩阵，可以通过OpenCV，MATLAB，ROS等方式得到
内容一般包含
Camera intrinsic matrix:
[708.0230464853036, 0, 313.299267971209;
0, 714.8509096655857, 189.0366118434282;
0, 0, 1]
Distorted arguments:
[-0.09135352304365538, 0.9392941836066392, -0.00487793090233253, -0.005752201943911559, -2.133389933308491]其中 3*3矩阵对应着 fx=708.023,fy=714.850,cx=313.2992,cy=189.0366
畸变系数k1=-0.0913, k2=0.9393, p1=-0.0049, p2=-0058。
如果使用的是webcam，也就是USB插入的摄像头，需要引入一个节点发布image消息，参考publisher_image
如果使用的是D415，可以在rs_camera.launch文件中对应的nodelet.launch.xml文件中增加remap
<node pkg="nodelet" type="nodelet" name="realsense2_camera" args="load realsense2_camera/RealSenseNodeFactory $(arg manager)">
<remap from="/camera/color/image_raw" to="/camera/image_raw" /> 再启动rs_camera.launch即可通过摄像头来使用ORB