与此物大战六个小时,终于解决所有问题,写此短文记录一下。
一、在github上下载源码后,解压缩,如下
进行g2o_with_orbslam2的编译
(1)报错信息:
ORBSLAM2_with_pointcloud_map/orbslam2_modified/g2o_with_orbslam2/g2o/types/slam2d/edge_se2_pointxy_bearing.cpp: In member function ‘virtual void g2o::EdgeSE2PointXYBearing::initialEstimate(const VertexSet&, g2o::OptimizableGraph::Vertex*)’:
/home/robot/g2o/g2o/types/slam2d/edge_se2_pointxy_bearing.cpp:50:52: error: no matching function for call to ‘g2o::SE2::setRotation(Eigen::Rotation2D::Scalar)’
t.setRotation(t.rotation().angle()+_measurement);
处理方法:
In g2o/types/slam2d/edge_se2_pointxy_bearing.cpp
change:
t.setRotation(t.rotation().angle()+_measurement);
to
t.setRotation((Eigen::Rotation2Dd)(t.rotation().angle()+_measurement));
(2)报错信息:
/usr/include/eigen3/Eigen/src/Core/util/StaticAssert.h:32: error: static assertion failed: YOU_MIXED_DIFFERENT_NUMERIC_TYPES__YOU_NEED_TO_USE_THE_CAST_METHOD_OF_MATRIXBASE_TO_CAST_NUMERIC_TYPES_EXPLICITLY
#define EIGEN_STATIC_ASSERT(X,MSG) static_assert(X,#MSG);
处理方法:
打开g2o/solvers/linear_solver_eigen.h,将以下代码:
template
class LinearSolverEigen: public LinearSolver
{
public:
typedef Eigen::SparseMatrix<double, Eigen::ColMajor> SparseMatrix;
typedef Eigen::Triplet Triplet;
typedef Eigen::PermutationMatrix<Eigen::Dynamic, Eigen::Dynamic, SparseMatrix::Index> PermutationMatrix;
改成
template
class LinearSolverEigen: public LinearSolver
{
public:
typedef Eigen::SparseMatrix<double, Eigen::ColMajor> SparseMatrix;
typedef Eigen::Triplet Triplet;
typedef Eigen::PermutationMatrix<Eigen::Dynamic, Eigen::Dynamic, int> PermutationMatrix;
(3)报错信息:
‘MatrixExponential’ is not a member of ‘Eigen’
处理方法:打开g2o/examples/target/continuous_to_discrete.h
// bigB = expm(bigA)
//Eigen::MatrixExponential me(bigA);
//me.compute(bigB);
bigB = bigA.exp();
前三行代码注释掉,添加最后一行
(4)报错信息:
ORBSLAM2_with_pointcloud_map/orbslam2_modified/g2o_with_orbslam2/g2o/examples/tutorial_slam2d/simulator.cpp:80:39: required from here
/usr/include/eigen3/Eigen/src/Core/util/StaticAssert.h:32:40: error: static assertion failed: FLOATING_POINT_ARGUMENT_PASSED__INTEGER_WAS_EXPECTED
#define EIGEN_STATIC_ASSERT(X,MSG) static_assert(X,#MSG);
处理方法:
打开g2o/g2o/examples/tutorial_slam2d/simulator.cpp:80行
VectorXd probLimits(MO_NUM_ELEMS);
改为
VectorXd probLimits; probLimits.resize(MO_NUM_ELEMS);
到此,g2o编译完成。
二、接着,进行源码编译,进入ORB_SLAM2_modified文件夹
我在直接执行build.sh时,出现bug(各种库版本问题,如opencv和pcl,按照我的方法可以解决这些问题),所以自行编译安装了
首先需要在Cmakelist。txt里把opencv版本要求注释掉,在添加位C++14编译,在第四行
由于build.sh始终编译不过,故放弃,选择自己编译
mkdir build
cd build
cmake ..
make
编译过成中出现诸多错误,一一列举出来
(1)g2o编译出错(在src下Optimizer.cc下有三处错误,错误原因一致,一一修改即可)
如error: no matching function for call to ‘g2o::BlockSolver<g2o::BlockSolverTraits<6, 3> >::BlockSolver(g2o::BlockSolver<g2o::BlockSolverTraits<6, 3> >::LinearSolverType*&)’
解决方法:
(I)
//g2o::BlockSolver_6_3 * solver_ptr = new g2o::BlockSolver_6_3(linearSolver); // line 356
g2o::BlockSolver_6_3 * solver_ptr = new g2o::BlockSolver_6_3(std::unique_ptr<g2o::BlockSolver_6_3::LinearSolverType> (linearSolver));
//g2o::OptimizationAlgorithmLevenberg * solver = new g2o::OptimizationAlgorithmLevenberg(solver_ptr); // line 357
g2o::OptimizationAlgorithmLevenberg * solver = new g2o::OptimizationAlgorithmLevenberg(std::unique_ptr<g2o::BlockSolver_6_3> (solver_ptr));
(II)
//g2o::BlockSolver_7_3 * solver_ptr = new g2o::BlockSolver_7_3(linearSolver); // line 356
g2o::BlockSolver_7_3 * solver_ptr = new g2o::BlockSolver_7_3(std::unique_ptr<g2o::BlockSolver_7_3::LinearSolverType> (linearSolver));
//g2o::OptimizationAlgorithmLevenberg * solver = new g2o::OptimizationAlgorithmLevenberg(solver_ptr); // line 357
g2o::OptimizationAlgorithmLevenberg * solver = new g2o::OptimizationAlgorithmLevenberg(std::unique_ptr<g2o::BlockSolver_7_3> (solver_ptr));
(III)
//g2o::BlockSolverX * solver_ptr = new g2o::BlockSolverX(linearSolver);
g2o::BlockSolverX * solver_ptr = new g2o::BlockSolverX(std::unique_ptr<g2o::BlockSolverX::LinearSolverType> (linearSolver));
//g2o::OptimizationAlgorithmLevenberg* solver = new g2o::OptimizationAlgorithmLevenberg(solver_ptr);
g2o::OptimizationAlgorithmLevenberg* solver = new g2o::OptimizationAlgorithmLevenberg(std::unique_ptr<g2o::BlockSolverX> (solver_ptr));
(2)编译Example下的例程出错,根据提示可知
分别进入各个报错源文件,按照上图所示一一进行修改即可
到此,编译问题解决,可以直接运行代码看效果了
./Examples/RGB-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/TUM1.yaml data1/rgbd_dataset_freiburg1_xyz/ Examples/RGB-D/associations.txt
(1)发现viewer界面是黑的,由此,修改代码
在ORB_SLAM2_modified/include/Tracking.h添加
// Current Frame
Frame mCurrentFrame;
cv::Mat mImRGB; //添加这行
cv::Mat mImGray;
cv::Mat mImDepth;
在ORB_SLAM2_modified/src/Tracking.cc修改2处
cv::Mat Tracking::GrabImageRGBD(const cv::Mat &imRGB,const cv::Mat &imD, const double ×tamp)
{
mImRGB = imRGB;//添加这行
mImGray = imRGB;
mImDepth = imD;
第二处
// insert Key Frame into point cloud viewer
//mpPointCloudMapping->insertKeyFrame( pKF, this->mImGray, this->mImDepth );
mpPointCloudMapping->insertKeyFrame( pKF, this->mImRGB, this->mImDepth ); //修改地方
如果修改这些还不能显示,可以试着滑动鼠标滚轮,把点云缩小,即可显示
(2)保存彩色点云地图
修改ORB_SLAM2_modified/src/pointcloudmapping.cc,在其中调用 PCL 库的pcl::io::savePCDFileBinary函数就可以保存点云地图了
加入头文件
#include <pcl/io/pcd_io.h>
在 void PointCloudMapping::viewer() 函数中( 123 行附近)加入保存地图的命令,最后样式如下:
...
for ( size_t i=lastKeyframeSize; i<N ; i++ )
{
PointCloud::Ptr p = generatePointCloud( keyframes[i], colorImgs[i], depthImgs[i] );
*globalMap += *p;
}
pcl::io::savePCDFileBinary("vslam.pcd", *globalMap); // 只需要加入这一句
...
修改之后重新编译程序
并且安装相应的工具,就可以查看生成的文件
#安装
sudo apt-get install pcl-tools
#查看
pcl_viewer vslam.pcd
大功告成!!!!