data_association.h

Go to the documentation of this file.

/* +------------------------------------------------------------------------+
   |                     Mobile Robot Programming Toolkit (MRPT)            |
   |                          http://www.mrpt.org/                          |
   |                                                                        |
   | Copyright (c) 2005-2018, Individual contributors, see AUTHORS file     |
   | See: http://www.mrpt.org/Authors - All rights reserved.                |
   | Released under BSD License. See details in http://www.mrpt.org/License |
   +------------------------------------------------------------------------+ */
#ifndef data_association_H
#define data_association_H

#include <mrpt/poses/CPoint2DPDFGaussian.h>
#include <mrpt/poses/CPointPDFGaussian.h>
#include <mrpt/math/CMatrixTemplate.h>  // mrpt::math::CMatrixBool
#include <mrpt/typemeta/TEnumType.h>

namespace mrpt::slam
{
/** \addtogroup data_assoc_grp Data association
 * \ingroup mrpt_slam_grp
 *  @{ */

/** @name Data association
    @{
  */

/** Different algorithms for data association, used in
 * mrpt::slam::data_association
 */
enum TDataAssociationMethod
{
    /** Nearest-neighbor. */
    assocNN = 0,
    /** JCBB: Joint Compatibility Branch & Bound [Neira, Tardos 2001]. */
    assocJCBB
};

/** Different metrics for data association, used in mrpt::slam::data_association
 *  For a comparison of both methods see paper:
 *  * J.L. Blanco, J. Gonzalez-Jimenez, J.A. Fernandez-Madrigal, "An
 * alternative to the Mahalanobis distance for determining optimal
 * correspondences in data association", IEEE Transactions on Robotics (T-RO),
 * (2012) DOI: 10.1109/TRO.2012.2193706 Draft:
 * http://ingmec.ual.es/~jlblanco/papers/blanco2012amd.pdf
 */
enum TDataAssociationMetric
{
    /** Mahalanobis distance */
    metricMaha = 0,
    /** Matching likelihood (See TDataAssociationMetric for a paper explaining
       this metric) */
    metricML
};

/** Used in mrpt::slam::TDataAssociationResults */
using observation_index_t = size_t;
/** Used in mrpt::slam::TDataAssociationResults */
using prediction_index_t = size_t;

/** The results from mrpt::slam::data_association
 */
struct TDataAssociationResults
{
    TDataAssociationResults()
        : associations(),
          distance(0),
          indiv_distances(0, 0),
          indiv_compatibility(0, 0),
          indiv_compatibility_counts(),
          nNodesExploredInJCBB(0)
    {
    }

    void clear()
    {
        associations.clear();
        distance = 0;
        indiv_distances.setSize(0, 0);
        indiv_compatibility.setSize(0, 0);
        indiv_compatibility_counts.clear();
        nNodesExploredInJCBB = 0;
    }

    /** For each observation (with row index IDX_obs in the input
     * "Z_observations"), its association in the predictions, as the row index
     * in the "Y_predictions_mean" input (or it's mapping to a custom ID, if it
     * was provided).
     * Note that not all observations may have an associated prediction.
     * An observation with index "IDX_obs" corresponds to the prediction number
     * "associations[IDX_obs]", or it may not correspond to anyone if it's not
     * present
     *  in the std::map (Tip: Use
     * associations.find(IDX_obs)!=associations.end() )
     * \note The types observation_index_t and prediction_index_t are just used
     * for clarity, use normal size_t's.
     */
    std::map<observation_index_t, prediction_index_t> associations;
    /** The Joint Mahalanobis distance or matching likelihood of the best
     * associations found. */
    double distance;

    /** Individual mahalanobis distances (or matching likelihood, depending on
     * the selected metric) between predictions (row indices) & observations
     * (column indices).
     *  Indices are for the appearing order in the arguments
     * "Y_predictions_mean" & "Z_observations", they are NOT landmark IDs.
     */
    mrpt::math::CMatrixDouble indiv_distances;
    /** The result of a chi2 test for compatibility using mahalanobis distance -
     * Indices are like in "indiv_distances". */
    mrpt::math::CMatrixBool indiv_compatibility;
    /** The sum of each column of indiv_compatibility, that is, the number of
     * compatible pairings for each observation. */
    std::vector<uint32_t> indiv_compatibility_counts;

    /** Only for the JCBB method,the number of recursive calls expent in the
     * algorithm. */
    size_t nNodesExploredInJCBB;
};

/** Computes the data-association between the prediction of a set of landmarks
 *and their observations, all of them with covariance matrices - Generic
 *version with prediction full cross-covariances.
 * Implemented methods include (see TDataAssociation)
 *      - NN: Nearest-neighbor
 *      - JCBB: Joint Compatibility Branch & Bound [Neira, Tardos 2001]
 *
 *  With both a Mahalanobis-distance or Matching-likelihood metric. For a
 *comparison of both methods, see paper:
 *  * J.L. Blanco, J. Gonzalez-Jimenez, J.A. Fernandez-Madrigal, "An
 *alternative to the Mahalanobis distance for determining optimal
 *correspondences in data association", IEEE Transactions on Robotics (T-RO),
 *(2012) DOI: 10.1109/TRO.2012.2193706 Draft:
 *http://ingmec.ual.es/~jlblanco/papers/blanco2012amd.pdf
 *
 * \param Z_observations_mean [IN] An MxO matrix with the M observations, each
 *row containing the observation "mean".
 * \param Y_predictions_mean [IN] An NxO matrix with the N predictions, each
 *row containing the mean of one prediction.
 * \param Y_predictions_cov [IN] An N*OxN*O matrix with the full covariance
 *matrix of all the N predictions.
 * \param results [OUT] The output data association hypothesis, and other
 *useful information.
 * \param method [IN, optional] The selected method to make the associations.
 * \param chi2quantile [IN, optional] The threshold for considering a match
 *between two close Gaussians for two landmarks, in the range [0,1]. It is used
 *to call mrpt::math::chi2inv
 * \param use_kd_tree [IN, optional] Build a KD-tree to speed-up the evaluation
 *of individual compatibility (IC). It's perhaps more efficient to disable it
 *for a small number of features. (default=true).
 * \param predictions_IDs [IN, optional] (default:none) An N-vector. If
 *provided, the resulting associations in "results.associations" will not
 *contain prediction indices "i", but "predictions_IDs[i]".
 *
 * \sa data_association_independent_predictions,
 *data_association_independent_2d_points,
 *data_association_independent_3d_points
 */
void data_association_full_covariance(
    const mrpt::math::CMatrixDouble& Z_observations_mean,
    const mrpt::math::CMatrixDouble& Y_predictions_mean,
    const mrpt::math::CMatrixDouble& Y_predictions_cov,
    TDataAssociationResults& results,
    const TDataAssociationMethod method = assocJCBB,
    const TDataAssociationMetric metric = metricMaha,
    const double chi2quantile = 0.99, const bool DAT_ASOC_USE_KDTREE = true,
    const std::vector<prediction_index_t>& predictions_IDs =
        std::vector<prediction_index_t>(),
    const TDataAssociationMetric compatibilityTestMetric = metricMaha,
    const double log_ML_compat_test_threshold = 0.0);

/** Computes the data-association between the prediction of a set of landmarks
 *and their observations, all of them with covariance matrices - Generic
 *version with NO prediction cross-covariances.
 * Implemented methods include (see TDataAssociation)
 *      - NN: Nearest-neighbor
 *      - JCBB: Joint Compatibility Branch & Bound [Neira, Tardos 2001]
 *
 *  With both a Mahalanobis-distance or Matching-likelihood metric. For a
 *comparison of both methods, see paper:
 *  * J.L. Blanco, J. Gonzalez-Jimenez, J.A. Fernandez-Madrigal, "An
 *alternative to the Mahalanobis distance for determining optimal
 *correspondences in data association", IEEE Transactions on Robotics (T-RO),
 *(2012) DOI: 10.1109/TRO.2012.2193706 Draft:
 *http://ingmec.ual.es/~jlblanco/papers/blanco2012amd.pdf
 *
 * \param Z_observations_mean [IN] An MxO matrix with the M observations, each
 *row containing the observation "mean".
 * \param Y_predictions_mean [IN] An NxO matrix with the N predictions, each
 *row containing the mean of one prediction.
 * \param Y_predictions_cov [IN] An N*OxO matrix: A vertical stack of N
 *covariance matrix, one for each of the N prediction.
 * \param results [OUT] The output data association hypothesis, and other
 *useful information.
 * \param method [IN, optional] The selected method to make the associations.
 * \param chi2quantile [IN, optional] The threshold for considering a match
 *between two close Gaussians for two landmarks, in the range [0,1]. It is used
 *to call mrpt::math::chi2inv
 * \param use_kd_tree [IN, optional] Build a KD-tree to speed-up the evaluation
 *of individual compatibility (IC). It's perhaps more efficient to disable it
 *for a small number of features. (default=true).
 * \param predictions_IDs [IN, optional] (default:none) An N-vector. If
 *provided, the resulting associations in "results.associations" will not
 *contain prediction indices "i", but "predictions_IDs[i]".
 *
 * \sa data_association_full_covariance,
 *data_association_independent_2d_points,
 *data_association_independent_3d_points
 */
void data_association_independent_predictions(
    const mrpt::math::CMatrixDouble& Z_observations_mean,
    const mrpt::math::CMatrixDouble& Y_predictions_mean,
    const mrpt::math::CMatrixDouble& Y_predictions_cov,
    TDataAssociationResults& results,
    const TDataAssociationMethod method = assocJCBB,
    const TDataAssociationMetric metric = metricMaha,
    const double chi2quantile = 0.99, const bool DAT_ASOC_USE_KDTREE = true,
    const std::vector<prediction_index_t>& predictions_IDs =
        std::vector<prediction_index_t>(),
    const TDataAssociationMetric compatibilityTestMetric = metricMaha,
    const double log_ML_compat_test_threshold = 0.0);

/** @} */

/** @} */  // end of grouping

}
MRPT_ENUM_TYPE_BEGIN(mrpt::slam::TDataAssociationMethod)
using namespace mrpt::slam;
MRPT_FILL_ENUM(assocNN);
MRPT_FILL_ENUM(assocJCBB);
MRPT_ENUM_TYPE_END()

MRPT_ENUM_TYPE_BEGIN(mrpt::slam::TDataAssociationMetric)
using namespace mrpt::slam;
MRPT_FILL_ENUM(metricMaha);
MRPT_FILL_ENUM(metricML);
MRPT_ENUM_TYPE_END()

#endif