检测到堆栈粉碎-来自未使用的功能

Question

我有一个函数，可以使用本征3.3.7 从C ++中的3个向量计算转换矩阵。在此函数中，有两行导致程序另一部分中的“ 检测到堆栈崩溃”错误，而不是函数本身。奇怪的是，即使我不调用该函数，也会发生错误。如果我注释掉这两行，则不会发生错误。怎么会这样？

//Calculate transformation matrix from sensor space to robot    
Eigen::Matrix4f GetSensor2Robot(Eigen::Vector3f & origin, 
    Eigen::Vector3f & xPoint, Eigen::Vector3f & yPoint){

    Eigen::Matrix4f transformation = Eigen::Matrix4f::Identity();         
    Eigen::Vector3f offset = origin;
    transformation.col(3).head<3>() = offset;

    Eigen::Vector3f xAxis = xPoint - origin;
    xAxis.normalize();
    Eigen::Vector3f yAxis = yPoint - origin;
    yAxis.normalize();

    //Make axis orthogonal
    float dot = xAxis.dot(yAxis);
    // Force the dot product of the two input vectors to
    // fall within the domain for inverse cosine, which
    // is -1 <= x <= 1. 
    dot = ( dot < -1.0 ? -1.0 : ( dot > 1.0 ? 1.0 : dot ) );

    //Differece to 90 degrees
    double angleDiff = acos(dot) - M_PI/2;

    Eigen::Vector3f zAxis = xAxis.cross(yAxis);

    Eigen::AngleAxis<float> axisAngleX(angleDiff/2, zAxis);
    //THE FOLLOWING LINE CAUSES THE ERROR
    xAxis = axisAngleX.toRotationMatrix()*xAxis;

    Eigen::AngleAxis<float> axisAngleY(-angleDiff/2, zAxis);
    //THE FOLLOWING LINE CAUSES THE ERROR
    yAxis = axisAngleY.toRotationMatrix()*yAxis;

    transformation.col(0).head<3>() = xAxis;
    transformation.col(1).head<3>() = yAxis;
    transformation.col(2).head<3>() = zAxis;

    Eigen::Matrix4f inverse = transformation.inverse();

    return inverse;
}

实际发生错误的部分使用点云库：

pcl::MovingLeastSquares<  pcl::PointXYZ,  pcl::PointNormal > mls;
mls.setSearchMethod(organizedNeighbour);
mls.setComputeNormals(true);

mls.setInputCloud(inputCloud);
mls.setIndices(indices);
mls.setPolynomialOrder(1);
mls.setSearchRadius(0.1);
mls.setSqrGaussParam(0.1*0.1);
mls.setNumberOfThreads(MAX_NUMBER_OF_THREADS);

pcl::PointCloud<pcl::PointNormal>::Ptr smoothedCloud(new 
    pcl::PointCloud<pcl::PointNormal>());

//during the following call, the error occurs
mls.process(*smoothedCloud);

我在装有Eigen 3.3.4的另一台计算机上尝试了相同的代码，但未发生错误。