这儿转载书籍《视觉SLAM十四讲从理论到实践 第2版》第6章的内容来学习非线性优化。在理解高斯牛顿法的时候需要注意如下理解。
#include <iostream>
#include <chrono>
#include <Eigen/Core>
#include <Eigen/Dense>
#include <g2o/stuff/sampler.h>
using namespace std;
using namespace Eigen;
int main(int argc, char **argv) {
double ar = 1.0, br = 2.0, cr = 1.0; // 真实参数值
double ae = 2.0, be = -1.0, ce = 5.0; // 估计参数值
int N = 100; // 数据点
double w_sigma = 1.0; // 噪声Sigma值
double inv_sigma = 1.0 / w_sigma; // 大于1的噪声可以让梯度下降慢一些
g2o::Sampler::seedRand();
vector<double> x_data, y_data; // 数据,用于测试的100组数据
for (int i = 0; i < N; i++) {
double x = i / 100.0;
x_data.push_back(x);
y_data.push_back(exp(ar * x * x + br * x + cr) + g2o::Sampler::gaussRand(0, 0.02));
}
// 开始Gauss-Newton迭代
int iterations = 100; // 迭代次数
double cost = 0, lastCost = 0; // 本次迭代的cost和上一次迭代的cost
chrono::steady_clock::time_point t1 = chrono::steady_clock::now();
for (int iter = 0; iter < iterations; iter++) {
Matrix3d H = Matrix3d::Zero(); // Hessian = J^T W^{-1} J in Gauss-Newton
Vector3d b = Vector3d::Zero(); // bias
cost = 0;
for (int i = 0; i < N; i++) {
double xi = x_data[i], yi = y_data[i]; // 第i个数据点
double error = yi - exp(ae * xi * xi + be * xi + ce);
Vector3d J; // 雅可比矩阵,fx的一阶导数,注意误差函数的x为向量[a, b, c],和这儿的xi没有关系。
J[0] = -xi * xi * exp(ae * xi * xi + be * xi + ce); // de/da
J[1] = -xi * exp(ae * xi * xi + be * xi + ce); // de/db
J[2] = -exp(ae * xi * xi + be * xi + ce); // de/dc
H += inv_sigma * inv_sigma * J * J.transpose();//海森矩阵H
b += -inv_sigma * inv_sigma * error * J;//error为误差函数的值
cost += error * error;
}
// 求解线性方程 Hx=b,dx为向量[a, b, c],即待优化参数。
Vector3d dx = H.ldlt().solve(b);
if (isnan(dx[0])) {
cout << "result is nan!" << endl;
break;
}
if (iter > 0 && cost >= lastCost) {
cout << "cost: " << cost << ">= last cost: " << lastCost << ", break." << endl;
break;
}
ae += dx[0];
be += dx[1];
ce += dx[2];
lastCost = cost;
cout << "total cost: " << cost << ", \t\tupdate: " << dx.transpose() <<
"\t\testimated params: " << ae << "," << be << "," << ce << endl;
}
chrono::steady_clock::time_point t2 = chrono::steady_clock::now();
chrono::duration</double><double> time_used = chrono::duration_cast<chrono::duration</double><double>>(t2 - t1);
cout << "solve time cost = " << time_used.count() << " seconds. " << endl;
cout << "estimated abc = " << ae << ", " << be << ", " << ce << endl;
return 0;
}
</double></chrono></iostream> |
#include <iostream>
#include <chrono>
#include <Eigen/Core>
#include <Eigen/Dense>
#include <g2o/stuff/sampler.h>
using namespace std;
using namespace Eigen;
int main(int argc, char **argv) {
double ar = 1.0, br = 2.0, cr = 1.0; // 真实参数值
double ae = 2.0, be = -1.0, ce = 5.0; // 估计参数值
int N = 100; // 数据点
double w_sigma = 1.0; // 噪声Sigma值
double inv_sigma = 1.0 / w_sigma; // 大于1的噪声可以让梯度下降慢一些
g2o::Sampler::seedRand();
vector<double> x_data, y_data; // 数据,用于测试的100组数据
for (int i = 0; i < N; i++) {
double x = i / 100.0;
x_data.push_back(x);
y_data.push_back(exp(ar * x * x + br * x + cr) + g2o::Sampler::gaussRand(0, 0.02));
}
// 开始Gauss-Newton迭代
int iterations = 100; // 迭代次数
double cost = 0, lastCost = 0; // 本次迭代的cost和上一次迭代的cost
chrono::steady_clock::time_point t1 = chrono::steady_clock::now();
for (int iter = 0; iter < iterations; iter++) {
Matrix3d H = Matrix3d::Zero(); // Hessian = J^T W^{-1} J in Gauss-Newton
Vector3d b = Vector3d::Zero(); // bias
cost = 0;
for (int i = 0; i < N; i++) {
double xi = x_data[i], yi = y_data[i]; // 第i个数据点
double error = yi - exp(ae * xi * xi + be * xi + ce);
Vector3d J; // 雅可比矩阵,fx的一阶导数,注意误差函数的x为向量[a, b, c],和这儿的xi没有关系。
J[0] = -xi * xi * exp(ae * xi * xi + be * xi + ce); // de/da
J[1] = -xi * exp(ae * xi * xi + be * xi + ce); // de/db
J[2] = -exp(ae * xi * xi + be * xi + ce); // de/dc
H += inv_sigma * inv_sigma * J * J.transpose();//海森矩阵H
b += -inv_sigma * inv_sigma * error * J;//error为误差函数的值
cost += error * error;
}
// 求解线性方程 Hx=b,dx为向量[a, b, c],即待优化参数。
Vector3d dx = H.ldlt().solve(b);
if (isnan(dx[0])) {
cout << "result is nan!" << endl;
break;
}
if (iter > 0 && cost >= lastCost) {
cout << "cost: " << cost << ">= last cost: " << lastCost << ", break." << endl;
break;
}
ae += dx[0];
be += dx[1];
ce += dx[2];
lastCost = cost;
cout << "total cost: " << cost << ", \t\tupdate: " << dx.transpose() <<
"\t\testimated params: " << ae << "," << be << "," << ce << endl;
}
chrono::steady_clock::time_point t2 = chrono::steady_clock::now();
chrono::duration</double><double> time_used = chrono::duration_cast<chrono::duration</double><double>>(t2 - t1);
cout << "solve time cost = " << time_used.count() << " seconds. " << endl;
cout << "estimated abc = " << ae << ", " << be << ", " << ce << endl;
return 0;
}
</double></chrono></iostream>
继续阅读“非线性优化”本作品采用知识共享署名 4.0 国际许可协议进行许可。