quaternion <-> euler, lots of fixes.
This commit is contained in:
@@ -13,29 +13,50 @@ namespace wr {
|
||||
namespace geom {
|
||||
|
||||
|
||||
/**
|
||||
* Quaternions encode rotations in three-dimensional space. While technically
|
||||
* a quaternion is comprised of a real element and a complex vector<3>, for
|
||||
* the purposes of this library, it is modeled as a floating point 4D vector.
|
||||
*
|
||||
* For information on the underlying vector type, see the documentation for
|
||||
* wr::geom::Vector.
|
||||
*
|
||||
* The constructors are primarily intended for intended operations; in practice,
|
||||
* the quaternionf and quaterniond functions are more useful for constructing
|
||||
* quaternions from vectors and angles.
|
||||
*
|
||||
* Like vectors, quaternions carry an internal tolerance value ε that is used for
|
||||
* floating point comparisons. The wr::math namespace contains the default values
|
||||
* used for this; generally, a tolerance of 0.0001 is considered appropriate for
|
||||
* the uses of this library. The tolerance can be explicitly set with the
|
||||
* setEpsilon method.
|
||||
*/
|
||||
template <typename T>
|
||||
class Quaternion {
|
||||
public:
|
||||
/**
|
||||
* The default Quaternion constructor returns an identity quaternion.
|
||||
*/
|
||||
Quaternion() : v(Vector<T, 3>()), w(1.0)
|
||||
Quaternion() : v(Vector<T, 3> {0.0, 0.0, 0.0}), w(1.0)
|
||||
{
|
||||
wr::math::DefaultEpsilon(this->eps);
|
||||
this->w = std::fmod(this->w, this->maxRotation);
|
||||
v.setEpsilon(this->eps);
|
||||
this->constrainAngle();
|
||||
};
|
||||
|
||||
|
||||
/**
|
||||
* A Quaternion may be initialised with a Vector<T, 3> axis of rotation
|
||||
* and an angle of rotation.
|
||||
* and an angle of rotation. This doesn't do the angle transforms to simplify
|
||||
* internal operations.
|
||||
* @param _axis A three-dimensional vector of the same type as the Quaternion.
|
||||
* @param _angle The angle of rotation about the axis of rotation.
|
||||
*/
|
||||
Quaternion(Vector<T, 3> _axis, T _angle) : v(_axis), w(_angle)
|
||||
{
|
||||
wr::math::DefaultEpsilon(this->eps);
|
||||
this->w = std::fmod(this->w, this->maxRotation);
|
||||
this->constrainAngle();
|
||||
v.setEpsilon(this->eps);
|
||||
};
|
||||
|
||||
/**
|
||||
@@ -48,9 +69,23 @@ public:
|
||||
w(vector[3])
|
||||
{
|
||||
wr::math::DefaultEpsilon(this->eps);
|
||||
this->w = std::fmod(this->w, this->m
|
||||
this->constrainAngle();
|
||||
v.setEpsilon(this->eps);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Set the comparison tolerance for this quaternion.
|
||||
* @param epsilon A tolerance value.
|
||||
*/
|
||||
void
|
||||
setEpsilon(T epsilon)
|
||||
{
|
||||
this->eps = epsilon;
|
||||
this->v.setEpsilon(epsilon);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Return the axis of rotation of this quaternion.
|
||||
* @return The axis of rotation of this quaternion.
|
||||
@@ -79,7 +114,7 @@ public:
|
||||
* @return A non-negative real number.
|
||||
*/
|
||||
T
|
||||
norm()
|
||||
norm() const
|
||||
{
|
||||
T n = 0;
|
||||
|
||||
@@ -92,28 +127,91 @@ public:
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Compute the conjugate of a quaternion.
|
||||
* @return The conjugate of this quaternion.
|
||||
*/
|
||||
Quaternion
|
||||
complexConj()
|
||||
conjugate() const
|
||||
{
|
||||
return Quaternion(Vector<T, 4> {this->v[0], this->v[1], this->v[2], this->w})
|
||||
return Quaternion(Vector<T, 4> {-this->v[0], -this->v[1], -this->v[2], this->w});
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Compute the inverse of a quaternion.
|
||||
* @return The inverse of this quaternion.
|
||||
*/
|
||||
Quaternion
|
||||
inverse() const
|
||||
{
|
||||
T _norm = this->norm();
|
||||
|
||||
return this->conjugate() / (_norm * _norm);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Determine whether this is a unit quaternion.
|
||||
* @return true if this is a unit quaternion.
|
||||
*/
|
||||
bool
|
||||
isUnitQuaternion() const
|
||||
{
|
||||
return wr::math::WithinTolerance(this->norm(), (T)1.0, this->eps);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Return the quaternion as a Vector<T, 4>, with the axis of rotation
|
||||
* followed by the angle of rotation.
|
||||
* @return A vector representation of the quaternion.
|
||||
*/
|
||||
Vector<T, 4>
|
||||
asVector()
|
||||
asVector() const
|
||||
{
|
||||
return Vector<T, 4> {this->v[0], this->v[1], this->v[2], this->w};
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Rotate vector v about this quaternion.
|
||||
* @param v The vector to be rotated.
|
||||
* @return The rotated vector.
|
||||
*/
|
||||
Vector<T, 3>
|
||||
rotate(Vector<T, 3> v) const
|
||||
{
|
||||
return (this->conjugate() * v * (*this)).axis();
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Return the Euler angles for this quaternion as a vector of
|
||||
* <yaw, pitch, roll>. Users of this function should watch out
|
||||
* for gimball lock.
|
||||
* @return A vector<T, 3> containing <yaw, pitch, roll>
|
||||
*/
|
||||
Vector<T, 3>
|
||||
euler() const
|
||||
{
|
||||
T yaw, pitch, roll;
|
||||
T a = this->w, a2 = a * a;
|
||||
T b = this->v[0], b2 = b * b;
|
||||
T c = this->v[1], c2 = c * c;
|
||||
T d = this->v[2], d2 = d * d;
|
||||
|
||||
yaw = std::atan2(2 * ((a*b) + (c * d)), a2 - b2 - c2 + d2);
|
||||
pitch = std::asin(2 * ((b*d) - (a*c)));
|
||||
roll = std::atan2(2 * ((a * d) + (b * c)), a2 + b2 - c2 - d2);
|
||||
|
||||
return Vector<T, 3> {yaw, pitch, roll};
|
||||
}
|
||||
|
||||
/**
|
||||
* Perform quaternion addition with another quaternion.
|
||||
* @param other The quaternion to be added with this one.
|
||||
* @return
|
||||
* @return The result of adding the two quaternions together.
|
||||
*/
|
||||
Quaternion
|
||||
operator+(const Quaternion<T> &other) const
|
||||
@@ -122,6 +220,11 @@ public:
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Perform quaternion subtraction with another quaternion.
|
||||
* @param other The quaternion to be subtracted from this one.
|
||||
* @return The result of subtracting the other quaternion from this one.
|
||||
*/
|
||||
Quaternion
|
||||
operator-(const Quaternion<T> &other) const
|
||||
{
|
||||
@@ -129,6 +232,49 @@ public:
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Perform scalar multiplication.
|
||||
* @param k The scaling value.
|
||||
* @return A scaled quaternion.
|
||||
*/
|
||||
Quaternion
|
||||
operator*(const T k) const
|
||||
{
|
||||
return Quaternion(this->v * k, this->w * k);
|
||||
}
|
||||
|
||||
|
||||
/** Perform scalar division.
|
||||
* @param k The scalar divisor.
|
||||
* @return A scaled quaternion.
|
||||
*/
|
||||
Quaternion
|
||||
operator/(const T k) const
|
||||
{
|
||||
return Quaternion(this->v / k, this->w / k);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Perform quaternion Hamilton multiplication with a three-
|
||||
* dimensional vector; this is done by treating the vector
|
||||
* as a pure quaternion (e.g. with an angle of rotation of 0).
|
||||
* @param vector The vector to multiply with this quaternion.
|
||||
* @return The Hamilton product of the quaternion and vector.
|
||||
*/
|
||||
Quaternion
|
||||
operator*(const Vector<T, 3> &vector) const
|
||||
{
|
||||
return Quaternion(vector * this->w + this->v.cross(vector),
|
||||
(T)0.0);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Perform quaternion Hamilton multiplication.
|
||||
* @param other The other quaternion to multiply with this one.
|
||||
* @result The Hamilton product of the two quaternions.
|
||||
*/
|
||||
Quaternion
|
||||
operator*(const Quaternion<T> &other) const
|
||||
{
|
||||
@@ -141,6 +287,11 @@ public:
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Perform quaternion equality checking.
|
||||
* @param other The quaternion to check equality against.
|
||||
* @return True if the two quaternions are equal within their tolerance.
|
||||
*/
|
||||
bool
|
||||
operator==(const Quaternion<T> &other) const
|
||||
{
|
||||
@@ -149,6 +300,11 @@ public:
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Perform quaternion inequality checking.
|
||||
* @param other The quaternion to check inequality against.
|
||||
* @return True if the two quaternions are unequal within their tolerance.
|
||||
*/
|
||||
bool
|
||||
operator!=(const Quaternion<T> &other) const
|
||||
{
|
||||
@@ -156,6 +312,13 @@ public:
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Support stream output of a quaternion in the form `a + <i, j, k>`.
|
||||
* TODO: improve the formatting.
|
||||
* @param outs An output stream
|
||||
* @param q A quaternion
|
||||
* @return The output stream
|
||||
*/
|
||||
friend std::ostream&
|
||||
operator<<(std::ostream& outs, const Quaternion<T>& q)
|
||||
{
|
||||
@@ -164,18 +327,128 @@ public:
|
||||
}
|
||||
|
||||
private:
|
||||
static constexpr T minRotation = -4 * M_PI;
|
||||
static constexpr T maxRotation = 4 * M_PI;
|
||||
|
||||
Vector<T, 3> v; // axis of rotation
|
||||
T w; // angle of rotation
|
||||
T eps;
|
||||
|
||||
void
|
||||
constrainAngle()
|
||||
{
|
||||
if (this->w < 0.0) {
|
||||
this->w = std::fmod(this->w, this->minRotation);
|
||||
}
|
||||
else {
|
||||
this->w = std::fmod(this->w, this->maxRotation);
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
|
||||
/**
|
||||
* Type aliases are provided for float and double quaternions.
|
||||
*/
|
||||
typedef Quaternion<float> Quaternionf;
|
||||
typedef Quaternion<double> Quaterniond;
|
||||
|
||||
|
||||
/**
|
||||
* Return a float quaternion scaled appropriately from a vector and angle,
|
||||
* e.g. angle = cos(angle / 2), axis.unitVector() * sin(angle / 2).
|
||||
* @param axis The axis of rotation.
|
||||
* @param angle The angle of rotation.
|
||||
* @return A quaternion.
|
||||
*/
|
||||
static Quaternionf
|
||||
quaternionf(Vector3f axis, float angle)
|
||||
{
|
||||
return Quaternionf(axis.unitVector() * std::sin(angle / 2.0),
|
||||
std::cos(angle / 2.0));
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Return a double quaternion scaled appropriately from a vector and angle,
|
||||
* e.g. angle = cos(angle / 2), axis.unitVector() * sin(angle / 2).
|
||||
* @param axis The axis of rotation.
|
||||
* @param angle The angle of rotation.
|
||||
* @return A quaternion.
|
||||
*/
|
||||
static Quaterniond
|
||||
quaterniond(Vector3d axis, double angle)
|
||||
{
|
||||
return Quaterniond(axis.unitVector() * std::sin(angle / 2.0),
|
||||
std::cos(angle / 2.0));
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Given a vector of Euler angles in ZYX sequence (e.g. yaw, pitch, roll),
|
||||
* return a quaternion.
|
||||
* @param euler A vector Euler angle in ZYX sequence.
|
||||
* @return A Quaternion representation of the orientation represented
|
||||
* by the Euler angles.
|
||||
*/
|
||||
static Quaternionf
|
||||
quaternionf_from_euler(Vector3f euler)
|
||||
{
|
||||
float x, y, z, w;
|
||||
euler = euler / 2.0;
|
||||
|
||||
float cos_yaw = std::cos(euler[0]);
|
||||
float cos_pitch = std::cos(euler[1]);
|
||||
float cos_roll = std::cos(euler[2]);
|
||||
float sin_yaw = std::sin(euler[0]);
|
||||
float sin_pitch = std::sin(euler[1]);
|
||||
float sin_roll = std::sin(euler[2]);
|
||||
|
||||
x = (sin_yaw * cos_pitch * cos_roll) + (cos_yaw * sin_pitch * sin_roll);
|
||||
y = (sin_yaw * cos_pitch * sin_roll) - (cos_yaw * sin_pitch * cos_roll);
|
||||
z = (cos_yaw * cos_pitch * sin_roll) + (sin_yaw * sin_pitch * cos_roll);
|
||||
w = (cos_yaw * cos_pitch * cos_roll) - (sin_yaw * sin_pitch * sin_roll);
|
||||
|
||||
return Quaternionf(Vector4f {x, y, z, w});
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Given a vector of Euler angles in ZYX sequence (e.g. yaw, pitch, roll),
|
||||
* return a quaternion.
|
||||
* @param euler A vector Euler angle in ZYX sequence.
|
||||
* @return A Quaternion representation of the orientation represented
|
||||
* by the Euler angles.
|
||||
*/
|
||||
static Quaterniond
|
||||
quaterniond_from_euler(Vector3d euler)
|
||||
{
|
||||
double x, y, z, w;
|
||||
euler = euler / 2.0;
|
||||
|
||||
double cos_yaw = std::cos(euler[0]);
|
||||
double cos_pitch = std::cos(euler[1]);
|
||||
double cos_roll = std::cos(euler[2]);
|
||||
double sin_yaw = std::sin(euler[0]);
|
||||
double sin_pitch = std::sin(euler[1]);
|
||||
double sin_roll = std::sin(euler[2]);
|
||||
|
||||
x = (sin_yaw * cos_pitch * cos_roll) + (cos_yaw * sin_pitch * sin_roll);
|
||||
y = (sin_yaw * cos_pitch * sin_roll) - (cos_yaw * sin_pitch * cos_roll);
|
||||
z = (cos_yaw * cos_pitch * sin_roll) + (sin_yaw * sin_pitch * cos_roll);
|
||||
w = (cos_yaw * cos_pitch * cos_roll) - (sin_yaw * sin_pitch * sin_roll);
|
||||
|
||||
return Quaterniond(Vector4d {x, y, z, w});
|
||||
}
|
||||
|
||||
|
||||
// Helpful references for understanding quaternions:
|
||||
// + "Intro to Quaternions" - https://www.youtube.com/watch?v=fKIss4EV6ME
|
||||
// 15 minutes into this video I had a more intuitive understanding.
|
||||
// + "Quaternions and Rotations" - http://graphics.stanford.edu/courses/cs348a-17-winter/Papers/quaternion.pdf
|
||||
// + "Understanding Quaternions" - http://www.chrobotics.com/library/understanding-quaternions
|
||||
|
||||
|
||||
} // namespace geom
|
||||
} // namespace wr
|
||||
|
||||
|
||||
@@ -12,6 +12,11 @@
|
||||
#include <wrmath/math.h>
|
||||
|
||||
|
||||
// This implementation is essentially a C++ translation of a Python library
|
||||
// I wrote for Coursera's "Linear Algebra for Machine Learning" course. Many
|
||||
// of the test vectors come from quiz questions in the class.
|
||||
|
||||
|
||||
namespace wr {
|
||||
namespace geom {
|
||||
|
||||
@@ -27,15 +32,17 @@ template <typename T, size_t N>
|
||||
class Vector {
|
||||
public:
|
||||
/**
|
||||
* The default constructor creates a zero vector for a given
|
||||
* The default constructor creates a unit vector for a given
|
||||
* type and size.
|
||||
*/
|
||||
Vector()
|
||||
{
|
||||
wr::math::DefaultEpsilon(this->epsilon);
|
||||
T unitLength = (T)1.0 / std::sqrt(N);
|
||||
for (size_t i = 0; i < N; i++) {
|
||||
this->arr[i] = 0.0;
|
||||
this->arr[i] = unitLength;
|
||||
}
|
||||
|
||||
wr::math::DefaultEpsilon(this->epsilon);
|
||||
}
|
||||
|
||||
/**
|
||||
|
||||
Reference in New Issue
Block a user