Template Class CovarianceEllipsoid< float, 3 >

• Defined in File covariance_ellipsoid.hpp

Class Documentation

template<>
class CovarianceEllipsoid<float, 3>

Specialisation of covariance ellipsoids in three dimensions.

This creates ellipsoids for 3x3 covariance matrices, i.e. the usual sort for mobile robot navigation. It uses one of the eigen solvers. Currently it seems (we have not checked it thorougly yet) to correlate the order of eigenvalue-eigenvector pairs with the index of the matrix, which is good!

Usage:

This class is also more conveniently typedef’d as CovarianceEllipsoid3f.

You can either instantiate directly through the constructor or by hand via the compute() method. Recomputing is also possible without having to construct an entirely new instance.

CovarianceEllipsoid3f ellipsoid(M); // OR
CovarianceEllipsoid3f ellipsoid_hand;
ellipsoid_hand.compute(M);

Once generated, the ellipse properties can easily be extracted.

Vector3f lengths = ellipsoid.lengths(); // axis lengths.

See also

CovarianceEllipsoid.

Public Functions

CovarianceEllipsoid()

Default constructor, use with compute().

This constructor does not do anything - you must use it together with the compute() function. Sometimes useful when using this class with containers.

CovarianceEllipsoid(const ecl::linear_algebra::Matrix3f &M, const bool sort = true)

Automatically generates ellipsoid information at construction.

Using the specified input covariance matrix, this constructor automatically generates (indirectly calls compute()) the full ellipse information, making it instantly ready for use.

Parameters

• M – : the input covariance matrix (must be symmetric positive definite).

• sort – : if true, ensure the resulting set is a right-hand configuration.

inline virtual ~CovarianceEllipsoid()

void compute(const ecl::linear_algebra::Matrix3f &M, const bool sort = true)

Computes the covariance ellipsoid for the specified matrix.

This can either be called indirectly via the constructor or directly via this function. In either case, it resets the object and generates/computes the resulting ellipse from the specified input covariance matrix. By default it will sort and normalise the resulting eigenvectors so that they form a right handed system. You can turn this off by setting the sort argument to false.

Parameters

• M – : the input covariance matrix (must be symmetric positive definite).

• sort – : if true, ensure the resulting set is a right-hand configuration.

inline const ecl::linear_algebra::Vector3f &lengths()

Returns the precomputed ellipsoid minor/major axis lengths.

Returns the precomputed ellipsoid minor/major axis lengths.

Returns

Vector2d : the minor/major axis lengths.

inline const ecl::linear_algebra::Matrix3f &axes()

Returns the ellipse axes/covariance eigen vectors.

This returns the ellipse axes (aka covariance eigen vectors) as columns of a matrix.

Returns

Matrix2d : the set of eigenvectors as column vectors in a matrix.