手眼標定學習記錄（更新）

一、Eigen中quaternion的構造函數爲
Quaternion (const Scalar &w, const Scalar &x, const Scalar &y, const Scalar &z)，注意w在前。然而在內部存儲時eigen將四元數的w放在最後
例如通過Eigen::Vector4d q = q_AB.coeffs();訪問時，q中的最後一個元素纔是w。

Eigen中用四元數表示向量的旋轉
Quaternion：

Eigen::Quaterniond q(2, 0, 1, -3);
std::cout << “This quaternion consists of a scalar ” << q.w() << ” and a vector ” << std::endl << q.vec() << std::endl;
q.normalize();
std::cout << “To represent rotation, we need to normalize it such that its length is ” << q.norm() << std::endl;
Eigen::Vector3d v(1, 2, -1);
Eigen::Quaterniond p;
p.w() = 0;
p.vec() = v;
Eigen::Quaterniond rotatedP = q * p * q.inverse();
Eigen::Vector3d rotatedV = rotatedP.vec();
std::cout << “We can now use it to rotate a vector ” << std::endl << v << ” to ” << std::endl << rotatedV << std::endl;

 

 

Eigen::Matrix3d R = q.toRotationMatrix(); // convert a quaternion to a 3x3 rotation matrix
std::cout << “Compare with the result using an rotation matrix ” << std::endl << R * v << std::endl;

 

 

Eigen::Quaterniond a = Eigen::Quterniond::Identity();
Eigen::Quaterniond b = Eigen::Quterniond::Identity();
Eigen::Quaterniond c; // Adding two quaternion as two 4x1 vectors is not supported by the EIgen API. That is, c = a + b is not allowed. We have to do this in a hard way
c.w() = a.w() + b.w();
c.x() = a.x() + b.x();
c.y() = a.y() + b.y();
c.z() = a.z() + b.z();

 

要輸出Eigen形式的四元數，並且由Eigen::Affine轉化，或者是由Eigen::matrix轉化方法
 

Eigen::Transform<double, 3, Eigen::Affine> resultAffine(result);//其中result的類型是Eigen::matrix.實現的是從Eigen::matrix到Eigen::Affine類型的轉換

Eigen::Quaternion<double> Target_Quaternion(Target_Pose.rotation());
    std::stringstream ss;
    ss<<Target_Quaternion.w()<<" "<<Target_Quaternion.x()<<" "<<Target_Quaternion.y()<< " "<<Target_Quaternion.z()<<std::endl;
    std::cerr << "the target quaternion eigen type w,x,y,z= "<<ss.str()<<std::endl;

或者利用tf::transformEigenToTf轉化爲TF形式的Transform再輸出

 

二，各種TF關係

tf::StamptedTransform(),對於參數，相當於

frame_id=father

child_frame_id=child

frame_id->child_frame_id

date form child_frame to  frame_id

the origin coordination of child_frame_id in frame_id

 tf::lookupTransform(),相當於

target_frame=father

source_frame=child

target_frame->source_frame

date from source_frame to target_frame

the origin coordination of source_frame in target_frame

---

solvepnp相當於

camera frame=father

world frame=child

camera frame->world frame

data from world frame to camera frame

the origin coorfination of world frame in camera frame

與相機的外參正好相反，外參需要的是將相機座標系下的點轉到世界座標系下，即data from camera frame to world frame

 

在傳入位姿信息時，需要傳入的是Eigen類型的Quaternion和position作爲姿態和位置信息，最好不要轉換爲tf類型的數據，會出現莫名的錯誤。

對於Eigen::Affine3d來說，除了上述直接得到Eigen::Quaternion之外，也可以通過

std::cerr << "the target translation eigen type x,y,z= "<<Target_Pose.translation().x()<<","<<Target_Pose.translation().y()<<","<<Target_Pose.translation().z()<<std::endl;

得到位姿矩陣的位置信息，返回其x,y,z值用於傳入Target_Pose.Position.x,y,z