系列文章目录
SLAM系列——第一讲 预备知识[2023.1]
SLAM系列——第二讲 初识SLAM[2023.1]
SLAM系列——第三讲 三维空间刚体运动[2023.1]
SLAM系列——第四讲 李群李代数[2023.1]
SLAM系列——第五讲 相机与图像[2023.1]
SLAM系列——第六讲 非线性优化[2023.1]
SLAM系列——第七讲 视觉里程计1[2023.1]
SLAM系列——第八讲 视觉里程计2[2023.1]
SLAM系列——第九讲 后端1[2023.1]
SLAM系列——第十讲 后端2[2023.1]
SLAM系列——第十一讲 回环检测[2023.1]
SLAM系列——第十二讲 建图[2023.1]
SLAM系列——第十三讲 实践:设计SLAM系统[2023.1]
SLAM系列——第十四讲 SLAM:现在与未来[2023.1]
文章目录
3.1 旋转矩阵
详见。
3.2 实践:Eigen
详见。
#include <iostream>
using namespace std;
#include <ctime>
// Eigen 核心部分
#include <Eigen/Core>
// 稠密矩阵的代数运算(逆,特征值等)
#include <Eigen/Dense>
using namespace Eigen;
#define MATRIX_SIZE 50
/****************************
* 本程序演示了 Eigen 基本类型的使用
****************************/
int main(int argc, char **argv) {
// Eigen 中所有向量和矩阵都是Eigen::Matrix,它是一个模板类。它的前三个参数为:数据类型,行,列
// 声明一个2*3的float矩阵
Matrix<float, 2, 3> matrix_23;
// 同时,Eigen 通过 typedef 提供了许多内置类型,不过底层仍是Eigen::Matrix
// 例如 Vector3d 实质上是 Eigen::Matrix<double, 3, 1>,即三维向量
Vector3d v_3d;
// 这是一样的
Matrix<float, 3, 1> vd_3d;
// Matrix3d 实质上是 Eigen::Matrix<double, 3, 3>
Matrix3d matrix_33 = Matrix3d::Zero(); //初始化为零
// 如果不确定矩阵大小,可以使用动态大小的矩阵
Matrix<double, Dynamic, Dynamic> matrix_dynamic;
// 更简单的
MatrixXd matrix_x;
// 这种类型还有很多,我们不一一列举
// 下面是对Eigen阵的操作
// 输入数据(初始化)
matrix_23 << 1, 2, 3, 4, 5, 6;
// 输出
cout << "matrix 2x3 from 1 to 6: \n" << matrix_23 << endl;
// 用()访问矩阵中的元素
cout << "print matrix 2x3: " << endl;
for (int i = 0; i < 2; i++) {
for (int j = 0; j < 3; j++) cout << matrix_23(i, j) << "\t";
cout << endl;
}
// 矩阵和向量相乘(实际上仍是矩阵和矩阵)
v_3d << 3, 2, 1;
vd_3d << 4, 5, 6;
// 但是在Eigen里你不能混合两种不同类型的矩阵,像这样是错的
// Matrix<double, 2, 1> result_wrong_type = matrix_23 * v_3d;
// 应该显式转换
Matrix<double, 2, 1> result = matrix_23.cast<double>() * v_3d;
cout << "[1,2,3;4,5,6]*[3,2,1]=" << result.transpose() << endl;
Matrix<float, 2, 1> result2 = matrix_23 * vd_3d;
cout << "[1,2,3;4,5,6]*[4,5,6]: " << result2.transpose() << endl;
// 同样你不能搞错矩阵的维度
// 试着取消下面的注释,看看Eigen会报什么错
// Eigen::Matrix<double, 2, 3> result_wrong_dimension = matrix_23.cast<double>() * v_3d;
// 一些矩阵运算
// 四则运算就不演示了,直接用+-*/即可。
matrix_33 = Matrix3d::Random(); // 随机数矩阵
cout << "random matrix: \n" << matrix_33 << endl;
cout << "transpose: \n" << matrix_33.transpose() << endl; // 转置
cout << "sum: " << matrix_33.sum() << endl; // 各元素和
cout << "trace: " << matrix_33.trace() << endl; // 迹
cout << "times 10: \n" << 10 * matrix_33 << endl; // 数乘
cout << "inverse: \n" << matrix_33.inverse() << endl; // 逆
cout << "det: " << matrix_33.determinant() << endl; // 行列式
// 特征值
// 实对称矩阵可以保证对角化成功
SelfAdjointEigenSolver<Matrix3d> eigen_solver(matrix_33.transpose() * matrix_33);
cout << "Eigen values = \n" << eigen_solver.eigenvalues() << endl;
cout << "Eigen vectors = \n" << eigen_solver.eigenvectors() << endl;
// 解方程
// 我们求解 matrix_NN * x = v_Nd 这个方程
// N的大小在前边的宏里定义,它由随机数生成
// 直接求逆自然是最直接的,但是求逆运算量大
Matrix<double, MATRIX_SIZE, MATRIX_SIZE> matrix_NN
= MatrixXd::Random(MATRIX_SIZE, MATRIX_SIZE);
matrix_NN = matrix_NN * matrix_NN.transpose(); // 保证半正定
Matrix<double, MATRIX_SIZE, 1> v_Nd = MatrixXd::Random(MATRIX_SIZE, 1);
clock_t time_stt = clock(); // 计时
// 直接求逆
Matrix<double, MATRIX_SIZE, 1> x = matrix_NN.inverse() * v_Nd;
cout << "time of normal inverse is "
<< 1000 * (clock() - time_stt) / (double) CLOCKS_PER_SEC << "ms" << endl;
cout << "x = " << x.transpose() << endl;
// 通常用矩阵分解来求,例如QR分解,速度会快很多
time_stt = clock();
x = matrix_NN.colPivHouseholderQr().solve(v_Nd);
cout << "time of Qr decomposition is "
<< 1000 * (clock() - time_stt) / (double) CLOCKS_PER_SEC << "ms" << endl;
cout << "x = " << x.transpose() << endl;
// 对于正定矩阵,还可以用cholesky分解来解方程
time_stt = clock();
x = matrix_NN.ldlt().solve(v_Nd);
cout << "time of ldlt decomposition is "
<< 1000 * (clock() - time_stt) / (double) CLOCKS_PER_SEC << "ms" << endl;
cout << "x = " << x.transpose() << endl;
return 0;
}
终端输出:
/slambook2/ch3/cmake-build-debug/useEigen/eigenMatrix
matrix 2x3 from 1 to 6:
1 2 3
4 5 6
print matrix 2x3:
1 2 3
4 5 6
[1,2,3;4,5,6]*[3,2,1]=10 28
[1,2,3;4,5,6]*[4,5,6]: 32 77
random matrix:
0.680375 0.59688 -0.329554
-0.211234 0.823295 0.536459
0.566198 -0.604897 -0.444451
transpose:
0.680375 -0.211234 0.566198
0.59688 0.823295 -0.604897
-0.329554 0.536459 -0.444451
sum: 1.61307
trace: 1.05922
times 10:
6.80375 5.9688 -3.29554
-2.11234 8.23295 5.36459
5.66198 -6.04897 -4.44451
inverse:
-0.198521 2.22739 2.8357
1.00605 -0.555135 -1.41603
-1.62213 3.59308 3.28973
det: 0.208598
Eigen values =
0.0242899
0.992154
1.80558
Eigen vectors =
-0.549013 -0.735943 0.396198
0.253452 -0.598296 -0.760134
-0.796459 0.316906 -0.514998
time of normal inverse is 0.064ms
x = -55.7896 -298.793 130.113 -388.455 -159.312 160.654 -40.0416 -193.561 155.844 181.144 185.125 -62.7786 19.8333 -30.8772 -200.746 55.8385 -206.604 26.3559 -14.6789 122.719 -221.449 26.233 -318.95 -78.6931 50.1446 87.1986 -194.922 132.319 -171.78 -4.19736 11.876 -171.779 48.3047 84.1812 -104.958 -47.2103 -57.4502 -48.9477 -19.4237 28.9419 111.421 92.1237 -288.248 -23.3478 -275.22 -292.062 -92.698 5.96847 -93.6244 109.734
time of Qr decomposition is 0.04ms
x = -55.7896 -298.793 130.113 -388.455 -159.312 160.654 -40.0416 -193.561 155.844 181.144 185.125 -62.7786 19.8333 -30.8772 -200.746 55.8385 -206.604 26.3559 -14.6789 122.719 -221.449 26.233 -318.95 -78.6931 50.1446 87.1986 -194.922 132.319 -171.78 -4.19736 11.876 -171.779 48.3047 84.1812 -104.958 -47.2103 -57.4502 -48.9477 -19.4237 28.9419 111.421 92.1237 -288.248 -23.3478 -275.22 -292.062 -92.698 5.96847 -93.6244 109.734
time of ldlt decomposition is 0.017ms
x = -55.7896 -298.793 130.113 -388.455 -159.312 160.654 -40.0416 -193.561 155.844 181.144 185.125 -62.7786 19.8333 -30.8772 -200.746 55.8385 -206.604 26.3559 -14.6789 122.719 -221.449 26.233 -318.95 -78.6931 50.1446 87.1986 -194.922 132.319 -171.78 -4.19736 11.876 -171.779 48.3047 84.1812 -104.958 -47.2103 -57.4502 -48.9477 -19.4237 28.9419 111.421 92.1237 -288.248 -23.3478 -275.22 -292.062 -92.698 5.96847 -93.6244 109.734
Process finished with exit code 0
3.3 旋转向量和欧拉角
详见。
3.4 四元数
详见。
3.5 相似、仿射、射影变换
详见。
3.6 实践:Eigen几何模块
详见。
#include <iostream>
#include <cmath>
using namespace std;
#include <Eigen/Core>
#include <Eigen/Geometry>
using namespace Eigen;
// 本程序演示了 Eigen 几何模块的使用方法
int main(int argc, char **argv) {
// Eigen/Geometry 模块提供了各种旋转和平移的表示
// 3D 旋转矩阵直接使用 Matrix3d 或 Matrix3f
Matrix3d rotation_matrix = Matrix3d::Identity();
// 旋转向量使用 AngleAxis, 它底层不直接是Matrix,但运算可以当作矩阵(因为重载了运算符)
AngleAxisd rotation_vector(M_PI / 4, Vector3d(0, 0, 1)); //沿 Z 轴旋转 45 度
cout.precision(3);
cout << "rotation matrix =\n" << rotation_vector.matrix() << endl; //用matrix()转换成矩阵
// 也可以直接赋值
rotation_matrix = rotation_vector.toRotationMatrix();
// 用 AngleAxis 可以进行坐标变换
Vector3d v(1, 0, 0);
Vector3d v_rotated = rotation_vector * v;
cout << "(1,0,0) after rotation (by angle axis) = " << v_rotated.transpose() << endl;
// 或者用旋转矩阵
v_rotated = rotation_matrix * v;
cout << "(1,0,0) after rotation (by matrix) = " << v_rotated.transpose() << endl;
// 欧拉角: 可以将旋转矩阵直接转换成欧拉角
Vector3d euler_angles = rotation_matrix.eulerAngles(2, 1, 0); // ZYX顺序,即yaw-pitch-roll顺序
cout << "yaw pitch roll = " << euler_angles.transpose() << endl;
// 欧氏变换矩阵使用 Eigen::Isometry
Isometry3d T = Isometry3d::Identity(); // 虽然称为3d,实质上是4*4的矩阵
T.rotate(rotation_vector); // 按照rotation_vector进行旋转
T.pretranslate(Vector3d(1, 3, 4)); // 把平移向量设成(1,3,4)
cout << "Transform matrix = \n" << T.matrix() << endl;
// 用变换矩阵进行坐标变换
Vector3d v_transformed = T * v; // 相当于R*v+t
cout << "v tranformed = " << v_transformed.transpose() << endl;
// 对于仿射和射影变换,使用 Eigen::Affine3d 和 Eigen::Projective3d 即可,略
// 四元数
// 可以直接把AngleAxis赋值给四元数,反之亦然
Quaterniond q = Quaterniond(rotation_vector);
cout << "quaternion from rotation vector = " << q.coeffs().transpose()
<< endl; // 请注意coeffs的顺序是(x,y,z,w),w为实部,前三者为虚部
// 也可以把旋转矩阵赋给它
q = Quaterniond(rotation_matrix);
cout << "quaternion from rotation matrix = " << q.coeffs().transpose() << endl;
// 使用四元数旋转一个向量,使用重载的乘法即可
v_rotated = q * v; // 注意数学上是qvq^{-1}
cout << "(1,0,0) after rotation = " << v_rotated.transpose() << endl;
// 用常规向量乘法表示,则应该如下计算
cout << "should be equal to " << (q * Quaterniond(0, 1, 0, 0) * q.inverse()).coeffs().transpose() << endl;
return 0;
}
3.6.2 实际的坐标变换例子
#include <iostream>
#include <vector>
#include <algorithm>
#include <Eigen/Core>
#include <Eigen/Geometry>
using namespace std;
using namespace Eigen;
int main(int argc, char** argv) {
Quaterniond q1(0.35, 0.2, 0.3, 0.1), q2(-0.5, 0.4, -0.1, 0.2);
q1.normalize();
q2.normalize();
Vector3d t1(0.3, 0.1, 0.1), t2(-0.1, 0.5, 0.3);
Vector3d p1(0.5, 0, 0.2);
Isometry3d T1w(q1), T2w(q2);
T1w.pretranslate(t1);
T2w.pretranslate(t2);
Vector3d p2 = T2w * T1w.inverse() * p1;
cout << endl << p2.transpose() << endl;
return 0;
}
3.7 可视化演示
详见。
3.7.1 显示运动轨迹
#include <pangolin/pangolin.h>
#include <Eigen/Core>
#include <unistd.h>
// 本例演示了如何画出一个预先存储的轨迹
using namespace std;
using namespace Eigen;
// path to trajectory file
string trajectory_file = "./examples/trajectory.txt";
void DrawTrajectory(vector<Isometry3d, Eigen::aligned_allocator<Isometry3d>>);
int main(int argc, char **argv) {
vector<Isometry3d, Eigen::aligned_allocator<Isometry3d>> poses;
ifstream fin(trajectory_file);
if (!fin) {
cout << "cannot find trajectory file at " << trajectory_file << endl;
return 1;
}
while (!fin.eof()) {
double time, tx, ty, tz, qx, qy, qz, qw;
fin >> time >> tx >> ty >> tz >> qx >> qy >> qz >> qw;
Isometry3d Twr(Quaterniond(qw, qx, qy, qz));
Twr.pretranslate(Vector3d(tx, ty, tz));
poses.push_back(Twr);
}
cout << "read total " << poses.size() << " pose entries" << endl;
// draw trajectory in pangolin
DrawTrajectory(poses);
return 0;
}
/*******************************************************************************************/
void DrawTrajectory(vector<Isometry3d, Eigen::aligned_allocator<Isometry3d>> poses) {
// create pangolin window and plot the trajectory
pangolin::CreateWindowAndBind("Trajectory Viewer", 1024, 768);
glEnable(GL_DEPTH_TEST);
glEnable(GL_BLEND);
glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
pangolin::OpenGlRenderState s_cam(
pangolin::ProjectionMatrix(1024, 768, 500, 500, 512, 389, 0.1, 1000),
pangolin::ModelViewLookAt(0, -0.1, -1.8, 0, 0, 0, 0.0, -1.0, 0.0)
);
pangolin::View &d_cam = pangolin::CreateDisplay()
.SetBounds(0.0, 1.0, 0.0, 1.0, -1024.0f / 768.0f)
.SetHandler(new pangolin::Handler3D(s_cam));
while (pangolin::ShouldQuit() == false) {
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
d_cam.Activate(s_cam);
glClearColor(1.0f, 1.0f, 1.0f, 1.0f);
glLineWidth(2);
for (size_t i = 0; i < poses.size(); i++) {
// 画每个位姿的三个坐标轴
Vector3d Ow = poses[i].translation();
Vector3d Xw = poses[i] * (0.1 * Vector3d(1, 0, 0));
Vector3d Yw = poses[i] * (0.1 * Vector3d(0, 1, 0));
Vector3d Zw = poses[i] * (0.1 * Vector3d(0, 0, 1));
glBegin(GL_LINES);
glColor3f(1.0, 0.0, 0.0);
glVertex3d(Ow[0], Ow[1], Ow[2]);
glVertex3d(Xw[0], Xw[1], Xw[2]);
glColor3f(0.0, 1.0, 0.0);
glVertex3d(Ow[0], Ow[1], Ow[2]);
glVertex3d(Yw[0], Yw[1], Yw[2]);
glColor3f(0.0, 0.0, 1.0);
glVertex3d(Ow[0], Ow[1], Ow[2]);
glVertex3d(Zw[0], Zw[1], Zw[2]);
glEnd();
}
// 画出连线
for (size_t i = 0; i < poses.size(); i++) {
glColor3f(0.0, 0.0, 0.0);
glBegin(GL_LINES);
auto p1 = poses[i], p2 = poses[i + 1];
glVertex3d(p1.translation()[0], p1.translation()[1], p1.translation()[2]);
glVertex3d(p2.translation()[0], p2.translation()[1], p2.translation()[2]);
glEnd();
}
pangolin::FinishFrame();
usleep(5000); // sleep 5 ms
}
}
习题
详见。