Class CPose3DQuatPDFGaussian

Defined in File CPose3DQuatPDFGaussian.h

Inheritance Relationships

Base Type

public mrpt::poses::CPose3DQuatPDF (Class CPose3DQuatPDF)

Class Documentation

class CPose3DQuatPDFGaussian : public mrpt::poses::CPose3DQuatPDF 

Declares a class that represents a Probability Density function (PDF) of a 3D pose using a quaternion \( p(\mathbf{x}) = [x ~ y ~ z ~ qr ~ qx ~ qy ~ qz]^\top \).

This class implements that PDF using a mono-modal Gaussian distribution. See mrpt::poses::CPose3DQuatPDF for more details, or mrpt::poses::CPose3DPDF for classes based on Euler angles instead.

Uncertainty of pose composition operations ( \( y = x \oplus u \)) is implemented in the methods “CPose3DQuatPDFGaussian::operator+=” and “CPose3DQuatPDF::jacobiansPoseComposition”.

Note

Read also: “A tutorial on SE(3) transformation parameterizations and

on-manifold optimization”, in blanco_se3_tutorial

Public Functions

CPose3DQuatPDFGaussian(): Default constructor - set all values to zero.

CPose3DQuatPDFGaussian(mrpt::math::TConstructorFlags_Quaternions constructor_dummy_param): Constructor which left all the member uninitialized, for using when speed is critical - as argument, use UNINITIALIZED_QUATERNION.

explicit CPose3DQuatPDFGaussian(const CPose3DQuat &init_Mean): Constructor from a default mean value, covariance equals to zero.

CPose3DQuatPDFGaussian(const CPose3DQuat &init_Mean, const mrpt::math::CMatrixDouble77 &init_Cov): Constructor with mean and covariance.

explicit CPose3DQuatPDFGaussian(const CPosePDFGaussian &o): Constructor from a Gaussian 2D pose PDF (sets to 0 the missing variables).

explicit CPose3DQuatPDFGaussian(const CPose3DPDFGaussian &o): Constructor from an equivalent Gaussian in Euler angles representation.

inline const CPose3DQuat &getPoseMean() const

inline CPose3DQuat &getPoseMean()

void getMean(CPose3DQuat &mean_pose) const override

inline std::tuple<cov_mat_t, type_value> getCovarianceAndMean() const override

virtual void copyFrom(const CPose3DQuatPDF &o) override: Copy operator, translating if necessary (for example, between particles and gaussian representations)

void copyFrom(const CPosePDF &o): Copy operator, translating if necessary (for example, between particles and gaussian representations)

void copyFrom(const CPose3DPDFGaussian &o): Copy operator, translating if necessary (for example, between particles and gaussian representations)

bool saveToTextFile(const std::string &file) const override: Save the PDF to a text file, containing the 3D pose in the first line (x y z qr qx qy qz), then the covariance matrix in the next 7 lines.

void changeCoordinatesReference(const CPose3DQuat &newReferenceBase): this = p (+) this. This can be used to convert a PDF from local coordinates to global, providing the point (newReferenceBase) from which “to project” the current pdf. Result PDF substituted the currently stored one in the object.

virtual void changeCoordinatesReference(const CPose3D &newReferenceBase) override: this = p (+) this. This can be used to convert a PDF from local coordinates to global, providing the point (newReferenceBase) from which “to project” the current pdf. Result PDF substituted the currently stored one in the object.

void drawSingleSample(CPose3DQuat &outPart) const override: Draws a single sample from the distribution

void drawManySamples(size_t N, std::vector<mrpt::math::CVectorDouble> &outSamples) const override: Draws a number of samples from the distribution, and saves as a list of 1x7 vectors, where each row contains a (x,y,z,qr,qx,qy,qz) datum.

mrpt::math::CMatrixDouble77 inverseJacobian() const: Compute the inverse Jacobian

virtual void inverse(CPose3DQuatPDF &o) const override: Returns a new PDF such as: NEW_PDF = (0,0,0) - THIS_PDF

mrpt::poses::CPose3DQuatPDFGaussian inverseCompositionCrossCorrelation(const mrpt::poses::CPose3DQuatPDFGaussian &pose_to) const: Returns the displacement from the current pose (pose_from) to pose_to: displacement = - pose_from + pose_to It assumes that both poses are correlated via the direct generative model: pose_to = pose_from + displacement For a deeper explanation, check https://github.com/MRPT/mrpt/pull/1243

inline CPose3DQuatPDFGaussian operator-() const: Unary - operator, returns the PDF of the inverse pose.

void operator+=(const CPose3DQuat &Ap): Makes: thisPDF = thisPDF + Ap, where “+” is pose composition (both the mean, and the covariance matrix are updated).

void operator+=(const CPose3DQuatPDFGaussian &Ap): Makes: thisPDF = thisPDF + Ap, where “+” is pose composition (both the mean, and the covariance matrix are updated) (see formulas in jacobiansPoseComposition ).

void operator-=(const CPose3DQuatPDFGaussian &Ap)

Makes: thisPDF = thisPDF - Ap, where “-” is pose inverse composition (both the mean, and the covariance matrix are updated). This operation assumes statistical independence between the two variables. If you want to take into account the cross-correlation between pose_1 and pose_2, use: displacement = pose_1.inverseCompositionCrossCorrelation(pose_2) Note: Both poses are correlated if the direct generative model is pose_2 = pose_1 + displacement