欧拉到四元数/四元数到欧拉使用特征

Euler to Quaternion / Quaternion to Euler using Eigen

本文关键字:四元数 特征      更新时间:2023-10-16

我正在尝试实现一种功能,可以将欧拉角转换为四元数,并使用Eigen返回"YXZ"惯例。稍后,这应该用于让用户为您提供欧拉角,并以四元数形式旋转并为用户转换回来。事实上,我的数学很差,但我已经尽力了。我不知道这个矩阵是否正确。代码工作,但我的结果是方式off,我想。你知道我哪里拐错弯了吗?这是我的Quat.cpp的样子:

#include "Quat.h"
#include <Eigen/Geometry>
#include <Eigen/Dense>
#include <cmath>
#include <iostream>
using namespace Eigen;
Vector3f Quat::MyRotation(const Vector3f YPR)
{
    Matrix3f matYaw(3, 3), matRoll(3, 3), matPitch(3, 3), matRotation(3, 3);
    const auto yaw = YPR[2]*M_PI / 180;
    const auto pitch = YPR[0]*M_PI / 180;
    const auto roll = YPR[1]*M_PI / 180;
    matYaw << cos(yaw), sin(yaw), 0.0f,
        -sin(yaw), cos(yaw), 0.0f,  //z
        0.0f, 0.0f, 1.0f;
    matPitch << cos(pitch), 0.0f, -sin(pitch),
        0.0f, 1.0f, 0.0f,   // X
        sin(pitch), 0.0f, cos(pitch);
    matRoll << 1.0f, 0.0f, 0.0f,
        0.0f, cos(roll), sin(roll),   // Y
        0.0f, -sin(roll), cos(roll);
    matRotation = matYaw*matPitch*matRoll;
    Quaternionf quatFromRot(matRotation);
    quatFromRot.normalize(); //Do i need to do this?
    return Quat::toYawPitchRoll(quatFromRot);
}
Vector3f Quat::toYawPitchRoll(const Eigen::Quaternionf& q)
{
    Vector3f retVector;
    const auto x = q.y();
    const auto y = q.z();
    const auto z = q.x();
    const auto w = q.w();
    retVector[2] = atan2(2.0 * (y * z + w * x), w * w - x * x - y * y + z * z);
    retVector[1] = asin(-2.0 * (x * z - w * y));
    retVector[0] = atan2(2.0 * (x * y + w * z), w * w + x * x - y * y - z * z);
#if 1
    retVector[0] = (retVector[0] * (180 / M_PI));
    retVector[1] = (retVector[1] * (180 / M_PI))*-1;
    retVector[2] = retVector[2] * (180 / M_PI);
#endif
    return retVector;
}

输入:x = 55.0, y = 80.0, z = 12.0四元数:w:0.872274, x: -0.140211, y:0.447012, z:-0.140211返回值:x:-55.5925, y: -6.84901, z:-21.8771忽略前缀,x值似乎是正确的,但Y和z是错误的。

从欧拉到四元数:

using namespace Eigen;
//Roll pitch and yaw in Radians
float roll = 1.5707, pitch = 0, yaw = 0.707;    
Quaternionf q;
q = AngleAxisf(roll, Vector3f::UnitX())
    * AngleAxisf(pitch, Vector3f::UnitY())
    * AngleAxisf(yaw, Vector3f::UnitZ());
std::cout << "Quaternion" << std::endl << q.coeffs() << std::endl;

从四元数到欧拉:

auto euler = q.toRotationMatrix().eulerAngles(0, 1, 2);
std::cout << "Euler from quaternion in roll, pitch, yaw"<< std::endl << euler << std::endl;

摘自https://eigen.tuxfamily.org/dox/classEigen_1_1AngleAxis.html

这里有一个方法(未测试):

  Vector3d euler = quaternion.toRotationMatrix().eulerAngles(2, 1, 0);
  yaw = euler[0]; pitch = euler[1]; roll = euler[2];

四元数到欧拉的解决方案不适合我,所以我研究和修改了代码,现在它为我的目的工作:

Vector3f ToEulerAngles(const Eigen::Quaternionf& q) {
    Vector3f angles;    //yaw pitch roll
    const auto x = q.x();
    const auto y = q.y();
    const auto z = q.z();
    const auto w = q.w();
    // roll (x-axis rotation)
    double sinr_cosp = 2 * (w * x + y * z);
    double cosr_cosp = 1 - 2 * (x * x + y * y);
    angles[2] = std::atan2(sinr_cosp, cosr_cosp);
    // pitch (y-axis rotation)
    double sinp = 2 * (w * y - z * x);
    if (std::abs(sinp) >= 1)
        angles[1] = std::copysign(M_PI / 2, sinp); // use 90 degrees if out of range
    else
        angles[1] = std::asin(sinp);
    // yaw (z-axis rotation)
    double siny_cosp = 2 * (w * z + x * y);
    double cosy_cosp = 1 - 2 * (y * y + z * z);
    angles[0] = std::atan2(siny_cosp, cosy_cosp);
    return angles;
}

我受到这个wiki条目的启发,并对这里提出的解决方案进行了一些基准测试。查看wiki:https://en.wikipedia.org/wiki/Conversion_between_quaternions_and_Euler_angles

当我使用

auto euler = q.toRotationMatrix()。eulerAngles(0, 1, 2)

它不可能一直完美工作,欧拉角总是有规律的跳动(实际值与计算值有±π的偏差)。例如,通过rqt读取和显示偏航角

我对此一无所知,但我发现ros tf::getYaw()也可以实现"四元数到欧拉"(因为我只需要偏航角)。

没有Eigen(以防万一),我这样做了:

tf2::Matrix3x3 ( quat ) . getEulerYPR( &roll, &pitch, &yaw );
// and
tf2::Matrix3x3 ( quat ) . getRPY( &roll, &pitch, &yaw );

但是,这些只能给出24种可能配置中的两种。