aloam框架laserMapping.cpp源码解读

news2024/11/18 10:47:18

一、详细源码解读

#include <math.h>
#include <vector>
#include <aloam_velodyne/common.h>
#include <nav_msgs/Odometry.h>
#include <nav_msgs/Path.h>
#include <geometry_msgs/PoseStamped.h>
#include <pcl_conversions/pcl_conversions.h>
#include <pcl/point_cloud.h>
#include <pcl/point_types.h>
#include <pcl/filters/voxel_grid.h>
#include <pcl/kdtree/kdtree_flann.h>
#include <ros/ros.h>
#include <sensor_msgs/Imu.h>
#include <sensor_msgs/PointCloud2.h>
#include <tf/transform_datatypes.h>
#include <tf/transform_broadcaster.h>
#include <eigen3/Eigen/Dense>
#include <ceres/ceres.h>
#include <mutex>
#include <queue>
#include <thread>
#include <iostream>
#include <string>

#include "lidarFactor.hpp"
#include "aloam_velodyne/common.h"
#include "aloam_velodyne/tic_toc.h"


int frameCount = 0;

double timeLaserCloudCornerLast = 0;
double timeLaserCloudSurfLast = 0;
double timeLaserCloudFullRes = 0;
double timeLaserOdometry = 0;


int laserCloudCenWidth = 10;
int laserCloudCenHeight = 10;
int laserCloudCenDepth = 5;
const int laserCloudWidth = 21;
const int laserCloudHeight = 21;
const int laserCloudDepth = 11;


const int laserCloudNum = laserCloudWidth * laserCloudHeight * laserCloudDepth; //4851


int laserCloudValidInd[125];
int laserCloudSurroundInd[125];

// input: from odom
pcl::PointCloud<PointType>::Ptr laserCloudCornerLast(new pcl::PointCloud<PointType>());
pcl::PointCloud<PointType>::Ptr laserCloudSurfLast(new pcl::PointCloud<PointType>());

// ouput: all visualble cube points
pcl::PointCloud<PointType>::Ptr laserCloudSurround(new pcl::PointCloud<PointType>());

// surround points in map to build tree
pcl::PointCloud<PointType>::Ptr laserCloudCornerFromMap(new pcl::PointCloud<PointType>());
pcl::PointCloud<PointType>::Ptr laserCloudSurfFromMap(new pcl::PointCloud<PointType>());

//input & output: points in one frame. local --> global
pcl::PointCloud<PointType>::Ptr laserCloudFullRes(new pcl::PointCloud<PointType>());

// points in every cube
pcl::PointCloud<PointType>::Ptr laserCloudCornerArray[laserCloudNum];
pcl::PointCloud<PointType>::Ptr laserCloudSurfArray[laserCloudNum];

//kd-tree
pcl::KdTreeFLANN<PointType>::Ptr kdtreeCornerFromMap(new pcl::KdTreeFLANN<PointType>());
pcl::KdTreeFLANN<PointType>::Ptr kdtreeSurfFromMap(new pcl::KdTreeFLANN<PointType>());

double parameters[7] = {0, 0, 0, 1, 0, 0, 0};
Eigen::Map<Eigen::Quaterniond> q_w_curr(parameters);
Eigen::Map<Eigen::Vector3d> t_w_curr(parameters + 4);

// wmap_T_odom * odom_T_curr = wmap_T_curr;
// transformation between odom's world and map's world frame
Eigen::Quaterniond q_wmap_wodom(1, 0, 0, 0);
Eigen::Vector3d t_wmap_wodom(0, 0, 0);

Eigen::Quaterniond q_wodom_curr(1, 0, 0, 0);
Eigen::Vector3d t_wodom_curr(0, 0, 0);


std::queue<sensor_msgs::PointCloud2ConstPtr> cornerLastBuf;
std::queue<sensor_msgs::PointCloud2ConstPtr> surfLastBuf;
std::queue<sensor_msgs::PointCloud2ConstPtr> fullResBuf;
std::queue<nav_msgs::Odometry::ConstPtr> odometryBuf;
std::mutex mBuf;

pcl::VoxelGrid<PointType> downSizeFilterCorner;
pcl::VoxelGrid<PointType> downSizeFilterSurf;

std::vector<int> pointSearchInd;
std::vector<float> pointSearchSqDis;

PointType pointOri, pointSel;

ros::Publisher pubLaserCloudSurround, pubLaserCloudMap, pubLaserCloudFullRes, pubOdomAftMapped, pubOdomAftMappedHighFrec, pubLaserAfterMappedPath;

nav_msgs::Path laserAfterMappedPath;

// set initial guess

void pointAssociateTobeMapped(PointType const *const pi, PointType *const po)
{
    Eigen::Vector3d point_w(pi->x, pi->y, pi->z);
    Eigen::Vector3d point_curr = q_w_curr.inverse() * (point_w - t_w_curr);
    po->x = point_curr.x();
    po->y = point_curr.y();
    po->z = point_curr.z();
    po->intensity = pi->intensity;
}

void laserCloudCornerLastHandler(const sensor_msgs::PointCloud2ConstPtr &laserCloudCornerLast2)
{
    mBuf.lock();
    cornerLastBuf.push(laserCloudCornerLast2);
    mBuf.unlock();
}

void laserCloudSurfLastHandler(const sensor_msgs::PointCloud2ConstPtr &laserCloudSurfLast2)
{
    mBuf.lock();
    surfLastBuf.push(laserCloudSurfLast2);
    mBuf.unlock();
}

void laserCloudFullResHandler(const sensor_msgs::PointCloud2ConstPtr &laserCloudFullRes2)
{
    mBuf.lock();
    fullResBuf.push(laserCloudFullRes2);
    mBuf.unlock();
}

//receive odomtry
void laserOdometryHandler(const nav_msgs::Odometry::ConstPtr &laserOdometry)
{
    mBuf.lock();
    odometryBuf.push(laserOdometry);
    mBuf.unlock();

    // high frequence publish
    Eigen::Quaterniond q_wodom_curr;
    Eigen::Vector3d t_wodom_curr;
    q_wodom_curr.x() = laserOdometry->pose.pose.orientation.x;
    q_wodom_curr.y() = laserOdometry->pose.pose.orientation.y;
    q_wodom_curr.z() = laserOdometry->pose.pose.orientation.z;
    q_wodom_curr.w() = laserOdometry->pose.pose.orientation.w;
    t_wodom_curr.x() = laserOdometry->pose.pose.position.x;
    t_wodom_curr.y() = laserOdometry->pose.pose.position.y;
    t_wodom_curr.z() = laserOdometry->pose.pose.position.z;

    Eigen::Quaterniond q_w_curr = q_wmap_wodom * q_wodom_curr;
    Eigen::Vector3d t_w_curr = q_wmap_wodom * t_wodom_curr + t_wmap_wodom;

    nav_msgs::Odometry odomAftMapped;
    odomAftMapped.header.frame_id = "camera_init";
    odomAftMapped.child_frame_id = "/aft_mapped";
    odomAftMapped.header.stamp = laserOdometry->header.stamp;
    odomAftMapped.pose.pose.orientation.x = q_w_curr.x();
    odomAftMapped.pose.pose.orientation.y = q_w_curr.y();
    odomAftMapped.pose.pose.orientation.z = q_w_curr.z();
    odomAftMapped.pose.pose.orientation.w = q_w_curr.w();
    odomAftMapped.pose.pose.position.x = t_w_curr.x();
    odomAftMapped.pose.pose.position.y = t_w_curr.y();
    odomAftMapped.pose.pose.position.z = t_w_curr.z();
    pubOdomAftMappedHighFrec.publish(odomAftMapped);
}

void process()
{
    while(1)
    {
        while (!cornerLastBuf.empty() && !surfLastBuf.empty() &&
            !fullResBuf.empty() && !odometryBuf.empty())
        {
            mBuf.lock();
            while (!odometryBuf.empty() && odometryBuf.front()->header.stamp.toSec() < cornerLastBuf.front()->header.stamp.toSec())
                odometryBuf.pop();
            if (odometryBuf.empty())
            {
                mBuf.unlock();
                break;
            }

            while (!surfLastBuf.empty() && surfLastBuf.front()->header.stamp.toSec() < cornerLastBuf.front()->header.stamp.toSec())
                surfLastBuf.pop();
            if (surfLastBuf.empty())
            {
                mBuf.unlock();
                break;
            }

            while (!fullResBuf.empty() && fullResBuf.front()->header.stamp.toSec() < cornerLastBuf.front()->header.stamp.toSec())
                fullResBuf.pop();
            if (fullResBuf.empty())
            {
                mBuf.unlock();
                break;
            }//确认时间同步

            timeLaserCloudCornerLast = cornerLastBuf.front()->header.stamp.toSec();
            timeLaserCloudSurfLast = surfLastBuf.front()->header.stamp.toSec();
            timeLaserCloudFullRes = fullResBuf.front()->header.stamp.toSec();
            timeLaserOdometry = odometryBuf.front()->header.stamp.toSec();

            if (timeLaserCloudCornerLast != timeLaserOdometry ||
                timeLaserCloudSurfLast != timeLaserOdometry ||
                timeLaserCloudFullRes != timeLaserOdometry)
            {
                printf("time corner %f surf %f full %f odom %f \n", timeLaserCloudCornerLast, timeLaserCloudSurfLast, timeLaserCloudFullRes, timeLaserOdometry);
                printf("unsync messeage!");
                mBuf.unlock();
                break;
            }

            laserCloudCornerLast->clear();
            pcl::fromROSMsg(*cornerLastBuf.front(), *laserCloudCornerLast);
            cornerLastBuf.pop();

//将数据从std::queue<sensor_msgs::PointCloud2ConstPtr>转为pcl::PointCloud<PointType>::Ptr

            laserCloudSurfLast->clear();
            pcl::fromROSMsg(*surfLastBuf.front(), *laserCloudSurfLast);
            surfLastBuf.pop();

            laserCloudFullRes->clear();
            pcl::fromROSMsg(*fullResBuf.front(), *laserCloudFullRes);
            fullResBuf.pop();

            q_wodom_curr.x() = odometryBuf.front()->pose.pose.orientation.x;
            q_wodom_curr.y() = odometryBuf.front()->pose.pose.orientation.y;
            q_wodom_curr.z() = odometryBuf.front()->pose.pose.orientation.z;
            q_wodom_curr.w() = odometryBuf.front()->pose.pose.orientation.w;
            t_wodom_curr.x() = odometryBuf.front()->pose.pose.position.x;
            t_wodom_curr.y() = odometryBuf.front()->pose.pose.position.y;
            t_wodom_curr.z() = odometryBuf.front()->pose.pose.position.z;
            odometryBuf.pop();

            while(!cornerLastBuf.empty())
            {
                cornerLastBuf.pop();
                printf("drop lidar frame in mapping for real time performance \n");
            }

            mBuf.unlock();//以上都是读取数据的代码

            TicToc t_whole;

            transformAssociateToMap();//为了便于学习，将代码放置如下：

void transformAssociateToMap()
{
    q_w_curr = q_wmap_wodom * q_wodom_curr;
    t_w_curr = q_wmap_wodom * t_wodom_curr + t_wmap_wodom;
}
//Eigen::Quaterniond q_wmap_wodom(1, 0, 0, 0);
//Eigen::Vector3d t_wmap_wodom(0, 0, 0);初值

            TicToc t_shift;
            int centerCubeI = int((t_w_curr.x() + 25.0) / 50.0) + laserCloudCenWidth;//10
            int centerCubeJ = int((t_w_curr.y() + 25.0) / 50.0) + laserCloudCenHeight;//10
            int centerCubeK = int((t_w_curr.z() + 25.0) / 50.0) + laserCloudCenDepth;//5

这里可以看LOAM笔记及A-LOAM源码阅读_a-loam github代码-CSDN博客

其中作者的图非常形象的展示了这个过程！感谢大佬！其中采用了基于空间范围划分的submap概念。

//将三维空间中的点映射到一个三维网格中，其中每个维度都被划分成了50单位大小的格子。

            if (t_w_curr.x() + 25.0 < 0)
                centerCubeI--;
            if (t_w_curr.y() + 25.0 < 0)
                centerCubeJ--;
            if (t_w_curr.z() + 25.0 < 0)
                centerCubeK--;

            while (centerCubeI < 3)//如果中心cube的值偏小，就将整体数组向小的方向移动
            {
                for (int j = 0; j < laserCloudHeight; j++)//21
                {
                    for (int k = 0; k < laserCloudDepth; k++)//11
                    {
                        int i = laserCloudWidth - 1;

//初始化一个变量 i，表示点云数据的宽度维度的索引，从最后一个元素开始
                        pcl::PointCloud<PointType>::Ptr laserCloudCubeCornerPointer =
                            laserCloudCornerArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];
                        pcl::PointCloud<PointType>::Ptr laserCloudCubeSurfPointer =
                            laserCloudSurfArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];
                        for (; i >= 1; i--)//从 i 开始递减，直到1。
                        {
                            laserCloudCornerArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] = laserCloudCornerArray[i - 1 + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];//更新角点云数据数组。
                            laserCloudSurfArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] = laserCloudSurfArray[i - 1 + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];
                        }
                        laserCloudCornerArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] =laserCloudCubeCornerPointer;//将更新后的角点云数据指针赋值回数组。
                        laserCloudSurfArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] =laserCloudCubeSurfPointer;
                        laserCloudCubeCornerPointer->clear();
                        laserCloudCubeSurfPointer->clear();
                    }
                }

                centerCubeI++;
                laserCloudCenWidth++;
            }

            while (centerCubeI >= laserCloudWidth - 3))//反之如果中心cube的值偏大，就将整体数组向大的方向移动
            {
                for (int j = 0; j < laserCloudHeight; j++)
                {
                    for (int k = 0; k < laserCloudDepth; k++)
                    {
                        int i = 0;
                        pcl::PointCloud<PointType>::Ptr laserCloudCubeCornerPointer =
                            laserCloudCornerArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];
                        pcl::PointCloud<PointType>::Ptr laserCloudCubeSurfPointer =
                            laserCloudSurfArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];
                        for (; i < laserCloudWidth - 1; i++)
                        {
                            laserCloudCornerArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] =laserCloudCornerArray[i + 1 + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];
                            laserCloudSurfArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] =laserCloudSurfArray[i + 1 + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k];
                        }
                        laserCloudCornerArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] =laserCloudCubeCornerPointer;
                        laserCloudSurfArray[i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k] =laserCloudCubeSurfPointer;
                        laserCloudCubeCornerPointer->clear();
                        laserCloudCubeSurfPointer->clear();
                    }
                }

                centerCubeI--;
                laserCloudCenWidth--;
            }

            while (centerCubeJ < 3)
            {
              //对centerCubeJ的处理方法一样，代码省略。
            }

            while (centerCubeJ >= laserCloudHeight - 3)
            {
                ......
            }

            while (centerCubeK < 3)
            {
                ......
            }

            while (centerCubeK >= laserCloudDepth - 3)
            {
              ......//最终结果应该是构建的点云格网位于点云的中部位置

//所以以上代码的作用应该是将三维点云数据的中心映射到三维格网中
            }

            int laserCloudValidNum = 0;//用于存储有效点云数据的数量。
            int laserCloudSurroundNum = 0;//用于存储周围点云数据的数量。

            for (int i = centerCubeI - 2; i <= centerCubeI + 2; i++)
            {
                for (int j = centerCubeJ - 2; j <= centerCubeJ + 2; j++)
                {
                    for (int k = centerCubeK - 1; k <= centerCubeK + 1; k++)
                    {
                        if (i >= 0 && i < laserCloudWidth &&
                            j >= 0 && j < laserCloudHeight &&
                            k >= 0 && k < laserCloudDepth)
                        {
                            laserCloudValidInd[laserCloudValidNum] = i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k;
                            laserCloudValidNum++;
                            laserCloudSurroundInd[laserCloudSurroundNum] = i + laserCloudWidth * j + laserCloudWidth * laserCloudHeight * k;
                            laserCloudSurroundNum++;
                        }//建立中心点的相邻点索引

//laserCloudCornerArray是与中心点相邻几个格网中的角点。
                    }
                }
            }

            laserCloudCornerFromMap->clear();

pcl::PointCloud<PointType>::Ptr laserCloudCornerFromMap(new pcl::PointCloud<PointType>());//智能指针


            laserCloudSurfFromMap->clear();
            for (int i = 0; i < laserCloudValidNum; i++)
            {
                *laserCloudCornerFromMap += *laserCloudCornerArray[laserCloudValidInd[i]];
                *laserCloudSurfFromMap += *laserCloudSurfArray[laserCloudValidInd[i]];

//将相邻的角点和面点加入到laserCloudCornerFromMap和laserCloudSurfFromMap
            }
            int laserCloudCornerFromMapNum = laserCloudCornerFromMap->points.size();
            int laserCloudSurfFromMapNum = laserCloudSurfFromMap->points.size();


            pcl::PointCloud<PointType>::Ptr laserCloudCornerStack(new pcl::PointCloud<PointType>());
            downSizeFilterCorner.setInputCloud(laserCloudCornerLast);
            downSizeFilterCorner.filter(*laserCloudCornerStack);

//downSizeFilterCorner 滤波器首先使用 setInputCloud 方法加载输入点云，然后调用 filter 方法执行下采样操作。下采样后的结果存储在 laserCloudCornerStack 点云对象中，这个对象随后可以用于进一步的处理或分析。

体素网格滤波器（VoxelGrid）是一种常见的点云下采样方法，它将空间划分为一个三维网格，并在每个体素内部平均或近似点云数据，以减少点的数量。

            int laserCloudCornerStackNum = laserCloudCornerStack->points.size();

            pcl::PointCloud<PointType>::Ptr laserCloudSurfStack(new pcl::PointCloud<PointType>());
            downSizeFilterSurf.setInputCloud(laserCloudSurfLast);
            downSizeFilterSurf.filter(*laserCloudSurfStack);
            int laserCloudSurfStackNum = laserCloudSurfStack->points.size();

            printf("map prepare time %f ms\n", t_shift.toc());//以上是准备工作的代码
            printf("map corner num %d  surf num %d \n", laserCloudCornerFromMapNum, laserCloudSurfFromMapNum);
            if (laserCloudCornerFromMapNum > 10 && laserCloudSurfFromMapNum > 50)
            {//我们先看如果相邻的角点和边点不够会怎么样？

//以下着重看怎么构建损失函数。
                TicToc t_opt;
                TicToc t_tree;
                kdtreeCornerFromMap->setInputCloud(laserCloudCornerFromMap);
                kdtreeSurfFromMap->setInputCloud(laserCloudSurfFromMap);

//为角点和边点构建kd树。
                printf("build tree time %f ms \n", t_tree.toc());

                for (int iterCount = 0; iterCount < 2; iterCount++)//优化两次
                {
                    //ceres::LossFunction *loss_function = NULL;
                    ceres::LossFunction *loss_function = new ceres::HuberLoss(0.1);

//参数 0.1 是尺度参数，用于控制损失函数在多大程度上对离群点（outliers）进行惩罚。
                    ceres::LocalParameterization *q_parameterization =
                        new ceres::EigenQuaternionParameterization();
                    ceres::Problem::Options problem_options;

                    ceres::Problem problem(problem_options);
                    problem.AddParameterBlock(parameters, 4, q_parameterization);
                    problem.AddParameterBlock(parameters + 4, 3);

                    TicToc t_data;
                    int corner_num = 0;

                    for (int i = 0; i < laserCloudCornerStackNum; i++)
                    {
                        pointOri = laserCloudCornerStack->points[i];
                        //double sqrtDis = pointOri.x * pointOri.x + pointOri.y * pointOri.y + pointOri.z * pointOri.z;
                        pointAssociateToMap(&pointOri, &pointSel);

//将点 pointOri 从雷达坐标系转换到世界坐标系。转换后的点存储在 pointSel 中。
                        kdtreeCornerFromMap->nearestKSearch(pointSel, 5, pointSearchInd, pointSearchSqDis);

// 在submap的corner特征点（target）中，寻找距离当前帧corner特征点（source）最近的5个点

参考自 LOAM笔记及A-LOAM源码阅读_a-loam github代码-CSDN博客

//求线的法向量用的是PCA方法

                        if (pointSearchSqDis[4] < 1.0)
                        {
                            std::vector<Eigen::Vector3d> nearCorners;
                            Eigen::Vector3d center(0, 0, 0);
                            for (int j = 0; j < 5; j++)
                            {
                                Eigen::Vector3d tmp(laserCloudCornerFromMap->points[pointSearchInd[j]].x,
                                                    laserCloudCornerFromMap->points[pointSearchInd[j]].y,
                                                    laserCloudCornerFromMap->points[pointSearchInd[j]].z);
                                center = center + tmp;
                                nearCorners.push_back(tmp);
                            }
                            center = center / 5.0;//计算五个最临近点的中心。

//计算协方差矩阵。

                            Eigen::Matrix3d covMat = Eigen::Matrix3d::Zero();
                            for (int j = 0; j < 5; j++)
                            {
                                Eigen::Matrix<double, 3, 1> tmpZeroMean = nearCorners[j] - center;
                                covMat = covMat + tmpZeroMean * tmpZeroMean.transpose();
                            }

//计算协方差矩阵的特征值和特征向量

                            Eigen::SelfAdjointEigenSolver<Eigen::Matrix3d> saes(covMat);

                            // if is indeed line feature
                            // note Eigen library sort eigenvalues in increasing order
                            Eigen::Vector3d unit_direction = saes.eigenvectors().col(2);

// 如果5个点呈线状分布，最大的特征值对应的特征向量就是该线的方向向量

                            Eigen::Vector3d curr_point(pointOri.x, pointOri.y, pointOri.z);
                            if (saes.eigenvalues()[2] > 3 * saes.eigenvalues()[1])
                            {

// 如果最大的特征值 >> 其他特征值，则5个点确实呈线状分布，否则认为直线“不够直”

                                Eigen::Vector3d point_on_line = center;
                                Eigen::Vector3d point_a, point_b;
                                point_a = 0.1 * unit_direction + point_on_line;
                                point_b = -0.1 * unit_direction + point_on_line;

 // 从中心点沿着方向向量向两端移动0.1m，构造线上的两个点

                                ceres::CostFunction *cost_function = LidarEdgeFactor::Create(curr_point, point_a, point_b, 1.0);

//残差距离即点到线的距离

                                problem.AddResidualBlock(cost_function, loss_function, parameters, parameters + 4);
                                corner_num++;    
                            }                            
                        }
                        /*
                        else if(pointSearchSqDis[4] < 0.01 * sqrtDis)
                        {
                            Eigen::Vector3d center(0, 0, 0);
                            for (int j = 0; j < 5; j++)
                            {
                                Eigen::Vector3d tmp(laserCloudCornerFromMap->points[pointSearchInd[j]].x,
                                                    laserCloudCornerFromMap->points[pointSearchInd[j]].y,
                                                    laserCloudCornerFromMap->points[pointSearchInd[j]].z);
                                center = center + tmp;
                            }
                            center = center / 5.0;    
                            Eigen::Vector3d curr_point(pointOri.x, pointOri.y, pointOri.z);
                            ceres::CostFunction *cost_function = LidarDistanceFactor::Create(curr_point, center);
                            problem.AddResidualBlock(cost_function, loss_function, parameters, parameters + 4);
                        }
                        */
                    }

                    int surf_num = 0;
                    for (int i = 0; i < laserCloudSurfStackNum; i++)
                    {
                        pointOri = laserCloudSurfStack->points[i];
                        //double sqrtDis = pointOri.x * pointOri.x + pointOri.y * pointOri.y + pointOri.z * pointOri.z;
                        pointAssociateToMap(&pointOri, &pointSel);
                        kdtreeSurfFromMap->nearestKSearch(pointSel, 5, pointSearchInd, pointSearchSqDis);

                        Eigen::Matrix<double, 5, 3> matA0;
                        Eigen::Matrix<double, 5, 1> matB0 = -1 * Eigen::Matrix<double, 5, 1>::Ones();

  • // 假设平面不通过原点，则平面的一般方程为Ax + By + Cz + 1 = 0。

  • // 用上面的2个矩阵表示平面方程就是 matA0 * norm（A, B, C） = matB0。

  • B0是全是-1的列向量，A0是观测值5个点的xyz坐标，要求的是法向量norm。

                        if (pointSearchSqDis[4] < 1.0)
                        {
                            
                            for (int j = 0; j < 5; j++)
                            {
                                matA0(j, 0) = laserCloudSurfFromMap->points[pointSearchInd[j]].x;
                                matA0(j, 1) = laserCloudSurfFromMap->points[pointSearchInd[j]].y;
                                matA0(j, 2) = laserCloudSurfFromMap->points[pointSearchInd[j]].z;
                                //printf(" pts %f %f %f ", matA0(j, 0), matA0(j, 1), matA0(j, 2));
                            }
                            // find the norm of plane
                            Eigen::Vector3d norm = matA0.colPivHouseholderQr().solve(matB0);

//使用 Eigen 库的 colPivHouseholderQr() 方法，对矩阵 matA0 进行列主元Householder QR分解，然后使用 solve 方法求解线性方程组 matA0 * X = matB0，得到平面的法向量 norm。
                            double negative_OA_dot_norm = 1 / norm.norm();
                            norm.normalize();

//调用 normalize() 方法将法向量 norm 归一化，使其长度为1。

                            // Here n(pa, pb, pc) is unit norm of plane
                            bool planeValid = true;
                            for (int j = 0; j < 5; j++)
                            {
                                // if OX * n > 0.2, then plane is not fit well不够平
                                if (fabs(norm(0) * laserCloudSurfFromMap->points[pointSearchInd[j]].x +
                                         norm(1) * laserCloudSurfFromMap->points[pointSearchInd[j]].y +
                                         norm(2) * laserCloudSurfFromMap->points[pointSearchInd[j]].z + negative_OA_dot_norm) > 0.2)

//

  • 这个表达式根据平面方程 Ax+By+Cz+D=0Ax+By+Cz+D=0 计算第 j 个最近邻点到平面的距离。这里 (A,B,C)(A,B,C) 是法向量 norm 的各分量，(x,y,z)(x,y,z) 是点的坐标。
  • negative_OA_dot_norm 是常数项 DD，在之前的步骤中已经计算过，它是法向量的模长的倒数。
  • fabs 函数用来计算表达式的绝对值，因为距离不能是负数。

                                {
                                    planeValid = false;
                                    break;
                                }
                            }
                            Eigen::Vector3d curr_point(pointOri.x, pointOri.y, pointOri.z);
                            if (planeValid)
                            {//构造点到面的距离残差项

                                ceres::CostFunction *cost_function = LidarPlaneNormFactor::Create(curr_point, norm, negative_OA_dot_norm);
                                problem.AddResidualBlock(cost_function, loss_function, parameters, parameters + 4);
                                surf_num++;
                            }
                        }
                        /*
                        else if(pointSearchSqDis[4] < 0.01 * sqrtDis)
                        {
                            Eigen::Vector3d center(0, 0, 0);
                            for (int j = 0; j < 5; j++)
                            {
                                Eigen::Vector3d tmp(laserCloudSurfFromMap->points[pointSearchInd[j]].x,
                                                    laserCloudSurfFromMap->points[pointSearchInd[j]].y,
                                                    laserCloudSurfFromMap->points[pointSearchInd[j]].z);
                                center = center + tmp;
                            }
                            center = center / 5.0;    
                            Eigen::Vector3d curr_point(pointOri.x, pointOri.y, pointOri.z);
                            ceres::CostFunction *cost_function = LidarDistanceFactor::Create(curr_point, center);
                            problem.AddResidualBlock(cost_function, loss_function, parameters, parameters + 4);
                        }
                        */
                    }

                    //printf("corner num %d used corner num %d \n", laserCloudCornerStackNum, corner_num);
                    //printf("surf num %d used surf num %d \n", laserCloudSurfStackNum, surf_num);

                    printf("mapping data assosiation time %f ms \n", t_data.toc());

                    TicToc t_solver;//求解损失向量
                    ceres::Solver::Options options;
                    options.linear_solver_type = ceres::DENSE_QR;
                    options.max_num_iterations = 4;
                    options.minimizer_progress_to_stdout = false;
                    options.check_gradients = false;
                    options.gradient_check_relative_precision = 1e-4;
                    ceres::Solver::Summary summary;
                    ceres::Solve(options, &problem, &summary);
                    printf("mapping solver time %f ms \n", t_solver.toc());

                    //printf("time %f \n", timeLaserOdometry);
                    //printf("corner factor num %d surf factor num %d\n", corner_num, surf_num);
                    //printf("result q %f %f %f %f result t %f %f %f\n", parameters[3], parameters[0], parameters[1], parameters[2],
                    //       parameters[4], parameters[5], parameters[6]);
                }
                printf("mapping optimization time %f \n", t_opt.toc());
            }
            else
            {
                ROS_WARN("time Map corner and surf num are not enough");
            }
            transformUpdate();

void transformUpdate()
{
	q_wmap_wodom = q_w_curr * q_wodom_curr.inverse();
	t_wmap_wodom = t_w_curr - q_wmap_wodom * t_wodom_curr;
}

            TicToc t_add;
            for (int i = 0; i < laserCloudCornerStackNum; i++)
            {
                pointAssociateToMap(&laserCloudCornerStack->points[i], &pointSel);

void pointAssociateToMap(PointType const *const pi, PointType *const po)
{
	Eigen::Vector3d point_curr(pi->x, pi->y, pi->z);
	Eigen::Vector3d point_w = q_w_curr * point_curr + t_w_curr;
	po->x = point_w.x();
	po->y = point_w.y();
	po->z = point_w.z();
	po->intensity = pi->intensity;
	//po->intensity = 1.0;
}

                int cubeI = int((pointSel.x + 25.0) / 50.0) + laserCloudCenWidth;
                int cubeJ = int((pointSel.y + 25.0) / 50.0) + laserCloudCenHeight;
                int cubeK = int((pointSel.z + 25.0) / 50.0) + laserCloudCenDepth;

//将三维点云数据映射到三维格网中

                if (pointSel.x + 25.0 < 0)
                    cubeI--;
                if (pointSel.y + 25.0 < 0)
                    cubeJ--;
                if (pointSel.z + 25.0 < 0)
                    cubeK--;

                if (cubeI >= 0 && cubeI < laserCloudWidth &&
                    cubeJ >= 0 && cubeJ < laserCloudHeight &&
                    cubeK >= 0 && cubeK < laserCloudDepth)
                {
                    int cubeInd = cubeI + laserCloudWidth * cubeJ + laserCloudWidth * laserCloudHeight * cubeK;
                    laserCloudCornerArray[cubeInd]->push_back(pointSel);

//为laserCloudCornerArray赋值
                }
            }

            for (int i = 0; i < laserCloudSurfStackNum; i++)
            {
                pointAssociateToMap(&laserCloudSurfStack->points[i], &pointSel);

                int cubeI = int((pointSel.x + 25.0) / 50.0) + laserCloudCenWidth;
                int cubeJ = int((pointSel.y + 25.0) / 50.0) + laserCloudCenHeight;
                int cubeK = int((pointSel.z + 25.0) / 50.0) + laserCloudCenDepth;

                if (pointSel.x + 25.0 < 0)
                    cubeI--;
                if (pointSel.y + 25.0 < 0)
                    cubeJ--;
                if (pointSel.z + 25.0 < 0)
                    cubeK--;

                if (cubeI >= 0 && cubeI < laserCloudWidth &&
                    cubeJ >= 0 && cubeJ < laserCloudHeight &&
                    cubeK >= 0 && cubeK < laserCloudDepth)
                {
                    int cubeInd = cubeI + laserCloudWidth * cubeJ + laserCloudWidth * laserCloudHeight * cubeK;
                    laserCloudSurfArray[cubeInd]->push_back(pointSel);
                }
            }
            printf("add points time %f ms\n", t_add.toc());

            
            TicToc t_filter;
            for (int i = 0; i < laserCloudValidNum; i++)
            {
                int ind = laserCloudValidInd[i];

                pcl::PointCloud<PointType>::Ptr tmpCorner(new pcl::PointCloud<PointType>());
                downSizeFilterCorner.setInputCloud(laserCloudCornerArray[ind]);
                downSizeFilterCorner.filter(*tmpCorner);
                laserCloudCornerArray[ind] = tmpCorner;

                pcl::PointCloud<PointType>::Ptr tmpSurf(new pcl::PointCloud<PointType>());
                downSizeFilterSurf.setInputCloud(laserCloudSurfArray[ind]);
                downSizeFilterSurf.filter(*tmpSurf);
                laserCloudSurfArray[ind] = tmpSurf;
            }
            printf("filter time %f ms \n", t_filter.toc());
            
            TicToc t_pub;
            //publish surround map for every 5 frame
            if (frameCount % 5 == 0)

//这个条件判断意味着每处理5帧数据，就会执行一次里面的代码块。
            {
                laserCloudSurround->clear();

//清空 laserCloudSurround 点云对象，为新的数据发布做准备。
                for (int i = 0; i < laserCloudSurroundNum; i++)
                {
                    int ind = laserCloudSurroundInd[i];
                    *laserCloudSurround += *laserCloudCornerArray[ind];
                    *laserCloudSurround += *laserCloudSurfArray[ind];
                }

                sensor_msgs::PointCloud2 laserCloudSurround3;
                pcl::toROSMsg(*laserCloudSurround, laserCloudSurround3);
                laserCloudSurround3.header.stamp = ros::Time().fromSec(timeLaserOdometry);
                laserCloudSurround3.header.frame_id = "camera_init";
                pubLaserCloudSurround.publish(laserCloudSurround3);

//*laserCloudSurround += *laserCloudSurfArray[ind]; 这两行将角点云和表面点云数据添加到 laserCloudSurround 中。

将 laserCloudSurround 转换为ROS消息格式 sensor_msgs::PointCloud2。

设置消息的时间戳和坐标帧ID。

通过 pubLaserCloudSurround.publish(laserCloudSurround3); 发布点云数据。

            }

            if (frameCount % 20 == 0)
            {
                pcl::PointCloud<PointType> laserCloudMap;
                for (int i = 0; i < 4851; i++)
                {
                    laserCloudMap += *laserCloudCornerArray[i];
                    laserCloudMap += *laserCloudSurfArray[i];
                }
                sensor_msgs::PointCloud2 laserCloudMsg;
                pcl::toROSMsg(laserCloudMap, laserCloudMsg);
                laserCloudMsg.header.stamp = ros::Time().fromSec(timeLaserOdometry);
                laserCloudMsg.header.frame_id = "camera_init";
                pubLaserCloudMap.publish(laserCloudMsg);
            }

            int laserCloudFullResNum = laserCloudFullRes->points.size();
            for (int i = 0; i < laserCloudFullResNum; i++)
            {
                pointAssociateToMap(&laserCloudFullRes->points[i], &laserCloudFullRes->points[i]);
            }

            sensor_msgs::PointCloud2 laserCloudFullRes3;
            pcl::toROSMsg(*laserCloudFullRes, laserCloudFullRes3);
            laserCloudFullRes3.header.stamp = ros::Time().fromSec(timeLaserOdometry);
            laserCloudFullRes3.header.frame_id = "camera_init";
            pubLaserCloudFullRes.publish(laserCloudFullRes3);

            printf("mapping pub time %f ms \n", t_pub.toc());

            printf("whole mapping time %f ms +++++\n", t_whole.toc());

            nav_msgs::Odometry odomAftMapped;
            odomAftMapped.header.frame_id = "camera_init";
            odomAftMapped.child_frame_id = "/aft_mapped";
            odomAftMapped.header.stamp = ros::Time().fromSec(timeLaserOdometry);
            odomAftMapped.pose.pose.orientation.x = q_w_curr.x();
            odomAftMapped.pose.pose.orientation.y = q_w_curr.y();
            odomAftMapped.pose.pose.orientation.z = q_w_curr.z();
            odomAftMapped.pose.pose.orientation.w = q_w_curr.w();
            odomAftMapped.pose.pose.position.x = t_w_curr.x();
            odomAftMapped.pose.pose.position.y = t_w_curr.y();
            odomAftMapped.pose.pose.position.z = t_w_curr.z();
            pubOdomAftMapped.publish(odomAftMapped);

            geometry_msgs::PoseStamped laserAfterMappedPose;
            laserAfterMappedPose.header = odomAftMapped.header;
            laserAfterMappedPose.pose = odomAftMapped.pose.pose;
            laserAfterMappedPath.header.stamp = odomAftMapped.header.stamp;
            laserAfterMappedPath.header.frame_id = "camera_init";
            laserAfterMappedPath.poses.push_back(laserAfterMappedPose);
            pubLaserAfterMappedPath.publish(laserAfterMappedPath);

            static tf::TransformBroadcaster br;
            tf::Transform transform;
            tf::Quaternion q;
            transform.setOrigin(tf::Vector3(t_w_curr(0),
                                            t_w_curr(1),
                                            t_w_curr(2)));
            q.setW(q_w_curr.w());
            q.setX(q_w_curr.x());
            q.setY(q_w_curr.y());
            q.setZ(q_w_curr.z());
            transform.setRotation(q);
            br.sendTransform(tf::StampedTransform(transform, odomAftMapped.header.stamp, "camera_init", "/aft_mapped"));

            frameCount++;
        }
        std::chrono::milliseconds dura(2);
        std::this_thread::sleep_for(dura);
    }
}

int main(int argc, char **argv)
{
    ros::init(argc, argv, "laserMapping");
    ros::NodeHandle nh;//初始化ros结点

    float lineRes = 0;
    float planeRes = 0;
    nh.param<float>("mapping_line_resolution", lineRes, 0.4);
    nh.param<float>("mapping_plane_resolution", planeRes, 0.8);

//0.4和0.8是划分的格网尺寸
    printf("line resolution %f plane resolution %f \n", lineRes, planeRes);
    downSizeFilterCorner.setLeafSize(lineRes, lineRes,lineRes);0.4*0.4*0.4的三维格网。
    downSizeFilterSurf.setLeafSize(planeRes, planeRes, planeRes);

    ros::Subscriber subLaserCloudCornerLast = nh.subscribe<sensor_msgs::PointCloud2>("/laser_cloud_corner_last", 100, laserCloudCornerLastHandler);

//接收来自里程计的数据，调用回调函数。

用std::queue<sensor_msgs::PointCloud2ConstPtr> cornerLastBuf;存储数据

    ros::Subscriber subLaserCloudSurfLast = nh.subscribe<sensor_msgs::PointCloud2>("/laser_cloud_surf_last", 100, laserCloudSurfLastHandler);

    ros::Subscriber subLaserOdometry = nh.subscribe<nav_msgs::Odometry>("/laser_odom_to_init", 100, laserOdometryHandler);

    ros::Subscriber subLaserCloudFullRes = nh.subscribe<sensor_msgs::PointCloud2>("/velodyne_cloud_3", 100, laserCloudFullResHandler);

    pubLaserCloudSurround = nh.advertise<sensor_msgs::PointCloud2>("/laser_cloud_surround", 100);

    pubLaserCloudMap = nh.advertise<sensor_msgs::PointCloud2>("/laser_cloud_map", 100);

    pubLaserCloudFullRes = nh.advertise<sensor_msgs::PointCloud2>("/velodyne_cloud_registered", 100);

    pubOdomAftMapped = nh.advertise<nav_msgs::Odometry>("/aft_mapped_to_init", 100);

    pubOdomAftMappedHighFrec = nh.advertise<nav_msgs::Odometry>("/aft_mapped_to_init_high_frec", 100);

    pubLaserAfterMappedPath = nh.advertise<nav_msgs::Path>("/aft_mapped_path", 100);

//这几行用于发布数据，大概发布的是配准后的点云数据和里程计数据，具体的数据用途我们在功能函数中看。

    for (int i = 0; i < laserCloudNum; i++)
    {
        laserCloudCornerArray[i].reset(new pcl::PointCloud<PointType>());
        laserCloudSurfArray[i].reset(new pcl::PointCloud<PointType>());
    }

//reset方法用于重置智能指针，使其指向一个新的pcl::PointCloud<PointType>对象。

    std::thread mapping_process{process};

//然后来看process函数

    ros::spin();

    return 0;
}

本文来自互联网用户投稿，该文观点仅代表作者本人，不代表本站立场。本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如若转载，请注明出处：http://www.coloradmin.cn/o/2169486.html

如若内容造成侵权/违法违规/事实不符，请联系多彩编程网进行投诉反馈，一经查实，立即删除！

相关文章

动手学深度学习8.7. 通过时间反向传播-笔记练习（PyTorch）

动手学深度学习8.7. 通过时间反向传播-笔记练习（PyTorch）

本节课程地址&#xff1a;本节无视频 本节教材地址&#xff1a;8.7. 通过时间反向传播 — 动手学深度学习 2.0.0 documentation (d2l.ai) 本节开源代码&#xff1a;...>d2l-zh>pytorch>chapter_multilayer-perceptrons>bptt.ipynb 通过时间反向传播 到目前为止&…
阅读更多...
[通义灵码] IDE 插件实现企业知识库问答

[通义灵码] IDE 插件实现企业知识库问答

在2024杭州云栖大会上&#xff0c;随着通义大模型能力的全面提升&#xff0c;阿里云通义灵码迎来重磅升级&#xff0c;从一年前只能完成基础的辅助编程任务&#xff0c;进化到几句话就能完成需求理解、任务拆解、代码编写、修改BUG、测试等开发任务&#xff0c;最快几分钟可从0…
阅读更多...
XSS | DOM 型 XSS 攻击

XSS | DOM 型 XSS 攻击

关注这个漏洞的其他相关笔记&#xff1a;XSS 漏洞 - 学习手册-CSDN博客 0x01&#xff1a;DOM 型 XSS —— 理论篇 DOM 全称 Document Object Model&#xff0c;使用 DOM 可以使程序和脚本能够动态访问和更新文档的内容、结构及样式。 DOM 型 XSS 是一种特殊类型的反射型 XSS&…
阅读更多...
系统实现悬浮窗-菜单-悬浮按钮功能

系统实现悬浮窗-菜单-悬浮按钮功能

文章目录 需求&#xff1a;系统实现悬浮窗菜单功能或悬浮小球定制功能实际手机产品效果悬浮窗作用 一、实际应用场景二、应用上面实现功能思路Demo演示效果部分源码分析Service层View层View初始化view 添加到窗体悬浮球拖动重点代码&#xff1a; 三、系统上面实现功能思路系统服…
阅读更多...
秒懂Linux之信号

秒懂Linux之信号

目录 信号的基本概念 信号的处理方式 默认动作 自定义处理信号 忽略该信号 信号的产生方式 kill命令 键盘组合键 系统调用 软件条件 异常 信号产生的深层理解 core的功能 信号的阻塞 内核中的表示 sigset_t 信号集操作函数 sigprocmask sigpending …
阅读更多...
do while循环

do while循环

/while(条件) {满足条件执行的代码&#xff0c;循环体 } /* do 做 */ while (false) { Console.WriteLine(" while循环执行了"); } do { //循环体逻辑 Console.WriteLine("dowhile循环执行了"); } while (true); Console.ReadLine(); /* w…
阅读更多...
数据库索引：最左匹配原则——提升数据库的查询性能

数据库索引：最左匹配原则——提升数据库的查询性能

数据库索引&#xff1a;最左匹配原则——提升数据库的查询性能 1、核心要点2、实例3、建议 &#x1f496;The Begin&#x1f496;点点关注&#xff0c;收藏不迷路&#x1f496; 在数据库优化中&#xff0c;组合索引的使用深受最左匹配原则的影响。这一原则是提升查询效率的关键…
阅读更多...
详细分析Nginx中的proxy_pass 末尾斜杠

详细分析Nginx中的proxy_pass 末尾斜杠

目录 前言1. 基本知识2. Demo 前言 对于Nginx的讲解&#xff0c;更多推荐阅读&#xff1a; Nginx配置静态网页访问&#xff08;图文界面&#xff09;Nginx将https重定向为http进行访问的配置&#xff08;附Demo&#xff09;Nginx从入门到精通&#xff08;全&#xff09;详细分…
阅读更多...
[Java EE] TCP 协议

[Java EE] TCP 协议

Author&#xff1a;MTingle major:人工智能 Build your hopes like a tower! 文章目录 文章目录​​​​​​​ 一. TCP 协议 二. TCP 特性 1. 确认应答(ack) 2. 超时重传 3. 连接管理 三次握手 四次挥手 TCP状态 4 滑动窗口 5. 流量控制 6.拥塞控制 7. 延时应答 8.捎带应答 9…
阅读更多...
前端性能初探

前端性能初探

前端监控 提升稳定性&#xff0c;更快的发现异常&#xff0c;定位异常&#xff0c;解决异常&#xff0c;js错误&#xff0c;接口异常&#xff0c;资源异常&#xff0c;白屏等。 关注用户体验&#xff0c;建立性能规范&#xff0c;长期关注优化&#xff0c;页面性能&#xff0c…
阅读更多...
TopOn对话游戏魔客：2024移动游戏广告应如何突破？

TopOn对话游戏魔客：2024移动游戏广告应如何突破？

TopOn对话游戏魔客&#xff1a;2024移动游戏广告应如何突破&#xff1f; 近年来&#xff0c;游戏广告投放的成本日益走高&#xff0c;ROI如何回正&#xff0c;素材如何创新等问题困扰着每一个广告主。在隐私政策的实施下&#xff0c;广告投放难度也在不断升级。 据data.ai发布…
阅读更多...
MK米客方德SD NAND参考设计

MK米客方德SD NAND参考设计

一、电路设计 参考电路&#xff1a; R1~R5 (10K-100 kΩ)是上拉电阻&#xff0c;当SD NAND处于高阻抗模式时&#xff0c;保护CMD和DAT线免受总线浮动。 即使主机使用SD NAND SD模式下的1位模式&#xff0c;主机也应通过上拉电阻上拉所有的DATO-3线。 R6&#xff08;RCLK&…
阅读更多...
解决图片放大模糊

解决图片放大模糊

首先需要了解设备像素和CSS像素&#xff0c;CSS像素 是 Web 开发中的逻辑像素&#xff0c;设计者根据这些像素来布局页面。设备像素 是设备屏幕上的实际像素点数。 DPR 是 设备像素 和 CSS像素 的比率&#xff0c;所以进行缩放后&#xff0c;也需要对图片尺寸进行处理&#xf…
阅读更多...
【HarmonyOS】鸿蒙自定义TabLayout示例

【HarmonyOS】鸿蒙自定义TabLayout示例

【HarmonyOS】自定义TabLayout代码示例&#xff0c;通过 Scroll 锚点 Tab 布局&#xff0c;滚动条会自动滚动使选中的标签居中显示。 class MyTabItem {label: string "";positionX: number -1; // 当前位置width: number -1; // 当前宽度constructor(label: stri…
阅读更多...
OpenHarmony（鸿蒙南向）——平台驱动指南【HDMI】

OpenHarmony（鸿蒙南向）——平台驱动指南【HDMI】

往期知识点记录&#xff1a; 鸿蒙&#xff08;HarmonyOS&#xff09;应用层开发&#xff08;北向&#xff09;知识点汇总 鸿蒙&#xff08;OpenHarmony&#xff09;南向开发保姆级知识点汇总~ 持续更新中…… 概述 功能简介 HDMI&#xff08;High Definition Multimedia Int…
阅读更多...
VS Code设置合集

VS Code设置合集

目录 VS Code设置合集1、汉化2、VS Code自动报错3、VS Code右键没有Open In Default Browser4、VS Code设置颜色主题5、修改默认缩进字符 VS Code设置合集 1、汉化 点击插件 → 搜索chinese → 点击install&#xff0c; 同时按住ctrl shift P → 搜索>configure displ…
阅读更多...
架构师：消息队列的技术指南

架构师：消息队列的技术指南

1、简述 消息队列（Message Queue, MQ）是一种异步通信机制，允许系统的各个组件通过消息在彼此之间进行通信。消息队列通过解耦系统组件、缓冲高峰期请求和提高系统的可扩展性，成为分布式系统中不可或缺的一部分。 2、工作原理 消息队列的基本工作原理是生产者将消息发布到…
阅读更多...
Lesson08---string（4）类

Lesson08---string（4）类

Lesson08—string类&#xff08;4&#xff09; c第八章string类的实现 文章目录 Lesson08---string类&#xff08;4&#xff09;前言一、计算机是怎么储存文字的1. 在此之前先思考一个问题2.编码表2.1 ascll码2.2unicode码2.3UTF码2.4gbk码 二、实现一个简单的string1.构造函数…
阅读更多...
【LeetCode】每日一题 2024_9_21 边积分最高的节点（哈希）

【LeetCode】每日一题 2024_9_21 边积分最高的节点（哈希）

前言 每天和你一起刷 LeetCode 每日一题~ LeetCode 启动&#xff01; 题目&#xff1a;边积分最高的节点 代码与解题思路 func edgeScore(edges []int) (ans int) {// 直接维护哈希最大值即可mp : map[int]int{}for i, v : range edges {mp[v] i// 如果多个节点的 边积分 相…
阅读更多...
Flutter中使用FFI的方式链接C/C++的so库(harmonyos)

Flutter中使用FFI的方式链接C/C++的so库(harmonyos)

Flutter中使用FFI的方式链接C/C库&#xff08;harmonyos&#xff09; FFI plugin创建和so的配置FFI插件对so库的使用 FFI plugin创建和so的配置 首先我们可以根据下面的链接生成FFI plugin插件&#xff1a;开发FFI plugin插件 然后在主项目中pubspec.yaml 添加插件的依赖路径&…
阅读更多...
推荐文章
最新文章

Copyright @ 2022~2023