lio-sam算法整理＜3＞

上次有一些函数没有查细节，这次接着查看

第一个函数就是 deskewPoint

thisPoint = deskewPoint(&thisPoint, laserCloudIn->points[i].time);

这里面传入了当前的点以及时间戳.

做的事情如下

    PointType deskewPoint(PointType *point, double relTime)
    {
        if (deskewFlag == -1 || cloudInfo.imuAvailable == false)
            return *point;

        double pointTime = timeScanCur + relTime;

        float rotXCur, rotYCur, rotZCur;
        findRotation(pointTime, &rotXCur, &rotYCur, &rotZCur);

        float posXCur, posYCur, posZCur;
        findPosition(relTime, &posXCur, &posYCur, &posZCur);

        if (firstPointFlag == true)  //每次这个重置都会为true
        {
            transStartInverse = (pcl::getTransformation(posXCur, posYCur, posZCur, rotXCur, rotYCur, rotZCur)).inverse();
            firstPointFlag = false;
        }

        // transform points to start
        Eigen::Affine3f transFinal = pcl::getTransformation(posXCur, posYCur, posZCur, rotXCur, rotYCur, rotZCur);
        Eigen::Affine3f transBt = transStartInverse * transFinal;
        //相当于每次都在初始值的位置上做一次复合.

        PointType newPoint;
        newPoint.x = transBt(0,0) * point->x + transBt(0,1) * point->y + transBt(0,2) * point->z + transBt(0,3);
        newPoint.y = transBt(1,0) * point->x + transBt(1,1) * point->y + transBt(1,2) * point->z + transBt(1,3);
        newPoint.z = transBt(2,0) * point->x + transBt(2,1) * point->y + transBt(2,2) * point->z + transBt(2,3);
        newPoint.intensity = point->intensity;

        return newPoint;
    }

主要过程是根据当前的时间戳，利用imu的数据找到 当前的位置和欧拉角.
然后得到外参

pcl::getTransformation(posXCur, posYCur, posZCur, rotXCur, rotYCur, rotZCur)

有一点还不太确定，不知道这里根据imu得到的信息是真实的世界坐标系还是相对于出发点的，不过个人理解得到的地图应该是相对于真实的世界坐标系的。

个人觉得这个函数主要是根据时间得到pose信息,然后再把点变到一个坐标系下.

值得注意的是

   void findPosition(double relTime, float *posXCur, float *posYCur, float *posZCur)
    {
        *posXCur = 0; *posYCur = 0; *posZCur = 0;

        // If the sensor moves relatively slow, like walking speed, positional deskew seems to have little benefits. Thus code below is commented.

        // if (cloudInfo.odomAvailable == false || odomDeskewFlag == false)
        //     return;

        // float ratio = relTime / (timeScanEnd - timeScanCur);

        // *posXCur = ratio * odomIncreX;
        // *posYCur = ratio * odomIncreY;
        // *posZCur = ratio * odomIncreZ;
    }

也就是findPosition返回的值总是原点. 所以上面才会猜测得到的地图就是就是以出发点为坐标原点来算的，因此每一次都会在出发点的基础上做一次复合，

     // transform points to start
        Eigen::Affine3f transFinal = pcl::getTransformation(posXCur, posYCur, posZCur, rotXCur, rotYCur, rotZCur);
        Eigen::Affine3f transBt = transStartInverse * transFinal;

        PointType newPoint;
        newPoint.x = transBt(0,0) * point->x + transBt(0,1) * point->y + transBt(0,2) * point->z + transBt(0,3);
        newPoint.y = transBt(1,0) * point->x + transBt(1,1) * point->y + transBt(1,2) * point->z + transBt(1,3);
        newPoint.z = transBt(2,0) * point->x + transBt(2,1) * point->y + transBt(2,2) * point->z + transBt(2,3);
        newPoint.intensity = point->intensity;

中间的变换其实就是矩阵相乘.

---

接下来继续看其他的代码

featureExtraction.cpp

这里也是主要是一个叫做 FeatureExtraction的类.

开始就声明了一个 接收器，三个发布器

   ros::Subscriber subLaserCloudInfo;
    ros::Publisher pubLaserCloudInfo;
    ros::Publisher pubCornerPoints;
    ros::Publisher pubSurfacePoints;

其详细实例如下

   subLaserCloudInfo = nh.subscribe<lio_sam::cloud_info>("lio_sam/deskew/cloud_info", 1, &FeatureExtraction::laserCloudInfoHandler, this, ros::TransportHints().tcpNoDelay());
        pubLaserCloudInfo = nh.advertise<lio_sam::cloud_info> ("lio_sam/feature/cloud_info", 1);
        pubCornerPoints = nh.advertise<sensor_msgs::PointCloud2>("lio_sam/feature/cloud_corner", 1);
        pubSurfacePoints = nh.advertise<sensor_msgs::PointCloud2>("lio_sam/feature/cloud_surface", 1);

可以看到

• subLaserCloudinfo就是接收的在之前imageProject.cpp里面lio-sam发出的cloud_info
• 其他的三个发布器都是预计这里处理完发布出去供后面的代码进行使用的.

接下来看一下接收器的callback函数是如何处理之前lio-sam发布的点云的信息的.
代码的风格和之前 imageProject.cpp里写的一样，即在一个callback函数中把主要函数都写在了这里面去.

    void laserCloudInfoHandler(const lio_sam::cloud_infoConstPtr& msgIn)
    {
        cloudInfo = *msgIn; // new cloud info
        cloudHeader = msgIn->header; // new cloud header
        pcl::fromROSMsg(msgIn->cloud_deskewed, *extractedCloud); // new cloud for extraction

        calculateSmoothness();

        markOccludedPoints();

        extractFeatures();

        publishFeatureCloud();
    }

上面的前三行是把 lio-sam发布的消息分别把对应的内容给到了

cloudInfo, cloudHeader,extractedCloud

calculateSmoothness函数

   void calculateSmoothness()
    {
        int cloudSize = extractedCloud->points.size();
        for (int i = 5; i < cloudSize - 5; i++)
        {
            float diffRange = cloudInfo.pointRange[i-5] + cloudInfo.pointRange[i-4]
                            + cloudInfo.pointRange[i-3] + cloudInfo.pointRange[i-2]
                            + cloudInfo.pointRange[i-1] - cloudInfo.pointRange[i] * 10
                            + cloudInfo.pointRange[i+1] + cloudInfo.pointRange[i+2]
                            + cloudInfo.pointRange[i+3] + cloudInfo.pointRange[i+4]
                            + cloudInfo.pointRange[i+5];

            cloudCurvature[i] = diffRange*diffRange;//diffX * diffX + diffY * diffY + diffZ * diffZ;

            cloudNeighborPicked[i] = 0;
            cloudLabel[i] = 0;
            // cloudSmoothness for sorting
            cloudSmoothness[i].value = cloudCurvature[i];
            cloudSmoothness[i].ind = i;
        }
    }

代码里面写的是计算曲率，但是并没有看懂，回头想明白了再来补充.

markOccludedPoints函数

 void markOccludedPoints()
    {
        int cloudSize = extractedCloud->points.size();
        // mark occluded points and parallel beam points
        for (int i = 5; i < cloudSize - 6; ++i)
        {
            // occluded points
            float depth1 = cloudInfo.pointRange[i];
            float depth2 = cloudInfo.pointRange[i+1];
            int columnDiff = std::abs(int(cloudInfo.pointColInd[i+1] - cloudInfo.pointColInd[i]));

            if (columnDiff < 10){
                // 10 pixel diff in range image
                if (depth1 - depth2 > 0.3){
                    cloudNeighborPicked[i - 5] = 1;
                    cloudNeighborPicked[i - 4] = 1;
                    cloudNeighborPicked[i - 3] = 1;
                    cloudNeighborPicked[i - 2] = 1;
                    cloudNeighborPicked[i - 1] = 1;
                    cloudNeighborPicked[i] = 1;
                }else if (depth2 - depth1 > 0.3){
                    cloudNeighborPicked[i + 1] = 1;
                    cloudNeighborPicked[i + 2] = 1;
                    cloudNeighborPicked[i + 3] = 1;
                    cloudNeighborPicked[i + 4] = 1;
                    cloudNeighborPicked[i + 5] = 1;
                    cloudNeighborPicked[i + 6] = 1;
                }
            }
            // parallel beam
            float diff1 = std::abs(float(cloudInfo.pointRange[i-1] - cloudInfo.pointRange[i]));
            float diff2 = std::abs(float(cloudInfo.pointRange[i+1] - cloudInfo.pointRange[i]));

            if (diff1 > 0.02 * cloudInfo.pointRange[i] && diff2 > 0.02 * cloudInfo.pointRange[i])
                cloudNeighborPicked[i] = 1;
        }
    }

这个函数主要是把 cloudNeighborPicked 中满足一定条件的点mark. 为什么这样做?

最重要的 extractFeature函数

这个函数比较长，大致看了一下，就是为了找到哪些点是 surface,哪些点是corner.

publishFeaturesCloud函数

  void publishFeatureCloud()
    {
        // free cloud info memory
        freeCloudInfoMemory();
        // save newly extracted features
        cloudInfo.cloud_corner  = publishCloud(&pubCornerPoints,  cornerCloud,  cloudHeader.stamp, lidarFrame);
        cloudInfo.cloud_surface = publishCloud(&pubSurfacePoints, surfaceCloud, cloudHeader.stamp, lidarFrame);
        // publish to mapOptimization
        pubLaserCloudInfo.publish(cloudInfo);
    }

这个函数就比较简单了，就是把之前得到的cornerpoints, surfacepoints给发布出去.