调试LSD-SLAM和问题解决

　　

安装的环境是ROS indigo + Ubuntu 14.04

一、打开一个终端，输入指令：sudo apt-get install python-rosinstall，这是后续工作的前提。
在Home文件夹下，依次在终端运行如下指令
mkdir ~/rosbuild_ws
cd ~/rosbuild_ws
rosws init . /opt/ros/indigo
mkdir package_dir
rosws set ~/rosbuild_ws/package_dir -t .
echo "source ~/rosbuild_ws/setup.bash" >> ~/.bashrc
bash
cd package_dir
sudo apt-get install ros-indigo-libg2o ros-indigo-cv-bridge liblapack-dev libblas-dev freeglut3-dev libqglviewer-dev libsuitesparse-dev libx11-dev

git clone https://github.com/tum-vision/lsd_slam.git lsd_slam
rosmake lsd_slam

二 碰到问题与解决
在该步骤：2.3 openFabMap for large loop-closure detection [optional]

出现错误//usr/local/lib/libopencv_nonfree.so: undefined reference to `cv::ocl::integral(cv::ocl::oclMat const&, cv::ocl::oclMat&)' collect2: error: ld returned 1 exit status

解决该问题的两个关键链接
https://github.com/tum-vision/lsd_slam/issues/27
http://stackoverflow.com/questions/26553659/linking-opencv-nonfree-components-sift-features-and-ocl-specifically

其实就是重新编译和安装OpenCV，

将“cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local ..”
改为“cmake -D CMAKE_BUILD_TYPE=RELEASE -D WITH_OPENCL=OFF -D CMAKE_INSTALL_PREFIX=/usr/local ..”
然后再“sudo make install”


三、重新编译lsd-slam
回到上面编译lsd-slam的终端，运行：

rosmake lsd_slam

错误将消失。

四、参照readme的内容跑程序。

转载于:https://www.cnblogs.com/huicanlin/p/5274570.html