CRangeBearingKFSLAM2D.cpp

Go to the documentation of this file.

/* +------------------------------------------------------------------------+
   |                     Mobile Robot Programming Toolkit (MRPT)            |
   |                          http://www.mrpt.org/                          |
   |                                                                        |
   | Copyright (c) 2005-2017, 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 |
   +------------------------------------------------------------------------+ */

#include "slam-precomp.h"  // Precompiled headers

#include <mrpt/slam/CRangeBearingKFSLAM2D.h>
#include <mrpt/slam/data_association.h>
#include <mrpt/poses/CPosePDF.h>
#include <mrpt/poses/CPosePDFGaussian.h>
#include <mrpt/obs/CActionRobotMovement3D.h>
#include <mrpt/math/utils.h>
#include <mrpt/math/wrap2pi.h>
#include <mrpt/utils/CTicTac.h>
#include <mrpt/system/os.h>

#include <mrpt/opengl/stock_objects.h>
#include <mrpt/opengl/CSetOfObjects.h>
#include <mrpt/opengl/CEllipsoid.h>

using namespace mrpt::slam;
using namespace mrpt::maps;
using namespace mrpt::math;
using namespace mrpt::obs;
using namespace mrpt::poses;
using namespace mrpt::utils;
using namespace mrpt::system;
using namespace mrpt;
using namespace std;

#define STATS_EXPERIMENT 0
#define STATS_EXPERIMENT_ALSO_NC 1

/*---------------------------------------------------------------
                                                        Constructor
  ---------------------------------------------------------------*/
CRangeBearingKFSLAM2D::CRangeBearingKFSLAM2D()
        : options(), m_action(), m_SF(), m_IDs(), m_SFs()
{
        reset();
}

/*---------------------------------------------------------------
                                                        reset
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::reset()
{
        m_action.reset();
        m_SF.reset();
        m_IDs.clear();
        m_SFs.clear();

        // INIT KF STATE
        m_xkk.assign(3, 0);  // State: 3D pose (x,y,phi)

        // Initial cov:
        m_pkk.setSize(3, 3);
        m_pkk.zeros();
}

/*---------------------------------------------------------------
                                                        Destructor
  ---------------------------------------------------------------*/
CRangeBearingKFSLAM2D::~CRangeBearingKFSLAM2D() {}
/*---------------------------------------------------------------
                                                        getCurrentRobotPose
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::getCurrentRobotPose(
        CPosePDFGaussian& out_robotPose) const
{
        MRPT_START

        ASSERT_(m_xkk.size() >= 3);

        // Set 6D pose mean:
        out_robotPose.mean = CPose2D(m_xkk[0], m_xkk[1], m_xkk[2]);

        // and cov:
        CMatrixTemplateNumeric<kftype> COV(3, 3);
        m_pkk.extractMatrix(0, 0, COV);
        out_robotPose.cov = COV;

        MRPT_END
}

/*---------------------------------------------------------------
                                                        getCurrentState
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::getCurrentState(
        CPosePDFGaussian& out_robotPose,
        std::vector<TPoint2D>& out_landmarksPositions,
        std::map<unsigned int, CLandmark::TLandmarkID>& out_landmarkIDs,
        CVectorDouble& out_fullState, CMatrixDouble& out_fullCovariance) const
{
        MRPT_START

        ASSERT_(m_xkk.size() >= 3);

        // Set 6D pose mean:
        out_robotPose.mean = CPose2D(m_xkk[0], m_xkk[1], m_xkk[2]);

        // and cov:
        CMatrixTemplateNumeric<kftype> COV(3, 3);
        m_pkk.extractMatrix(0, 0, COV);
        out_robotPose.cov = COV;

        // Landmarks:
        ASSERT_(((m_xkk.size() - 3) % 2) == 0);
        size_t i, nLMs = (m_xkk.size() - 3) / 2;
        out_landmarksPositions.resize(nLMs);
        for (i = 0; i < nLMs; i++)
        {
                out_landmarksPositions[i].x = m_xkk[3 + i * 2 + 0];
                out_landmarksPositions[i].y = m_xkk[3 + i * 2 + 1];
        }  // end for i

        // IDs:
        out_landmarkIDs = m_IDs.getInverseMap();  // m_IDs_inverse;

        // Full state:
        out_fullState.resize(m_xkk.size());
        for (KFVector::Index i = 0; i < m_xkk.size(); i++)
                out_fullState[i] = m_xkk[i];

        // Full cov:
        out_fullCovariance = m_pkk;

        MRPT_END
}

/*---------------------------------------------------------------
                                                processActionObservation
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::processActionObservation(
        CActionCollection::Ptr& action, CSensoryFrame::Ptr& SF)
{
        MRPT_START

        m_action = action;
        m_SF = SF;

        // Sanity check:
        ASSERT_(m_IDs.size() == this->getNumberOfLandmarksInTheMap());

        // ===================================================================================================================
        // Here's the meat!: Call the main method for the KF algorithm, which will
        // call all the callback methods as required:
        // ===================================================================================================================
        runOneKalmanIteration();

        // =============================================================
        //  ADD TO SFs SEQUENCE
        // =============================================================
        if (options.create_simplemap)
        {
                CPosePDFGaussian::Ptr auxPosePDF =
                        mrpt::make_aligned_shared<CPosePDFGaussian>();
                getCurrentRobotPose(*auxPosePDF);
                m_SFs.insert(auxPosePDF, SF);
        }

        MRPT_END
}

/** Must return the action vector u.
  * \param out_u The action vector which will be passed to OnTransitionModel
  */
void CRangeBearingKFSLAM2D::OnGetAction(KFArray_ACT& u) const
{
        // Get odometry estimation:
        CActionRobotMovement2D::Ptr actMov2D =
                m_action->getBestMovementEstimation();
        CActionRobotMovement3D::Ptr actMov3D =
                m_action->getActionByClass<CActionRobotMovement3D>();
        if (actMov3D)
        {
                u[0] = actMov3D->poseChange.mean.x();
                u[1] = actMov3D->poseChange.mean.y();
                u[2] = actMov3D->poseChange.mean.yaw();
        }
        else if (actMov2D)
        {
                CPose2D estMovMean;
                actMov2D->poseChange->getMean(estMovMean);
                u[0] = estMovMean.x();
                u[1] = estMovMean.y();
                u[2] = estMovMean.phi();
        }
        else
        {
                // Left u as zeros
                for (size_t i = 0; i < 3; i++) u[i] = 0;
        }
}

/** This virtual function musts implement the prediction model of the Kalman
 * filter.
 */
void CRangeBearingKFSLAM2D::OnTransitionModel(
        const KFArray_ACT& u, KFArray_VEH& xv, bool& out_skipPrediction) const
{
        MRPT_START

        // Do not update the vehicle pose & its covariance until we have some
        // landmakrs in the map,
        // otherwise, we are imposing a lower bound to the best uncertainty from now
        // on:
        if (size_t(m_xkk.size()) == get_vehicle_size())
        {
                out_skipPrediction = true;
                return;
        }

        CPose2D robotPose(xv[0], xv[1], xv[2]);
        CPose2D odoIncrement(u[0], u[1], u[2]);

        // Pose composition:
        robotPose = robotPose + odoIncrement;

        xv[0] = robotPose.x();
        xv[1] = robotPose.y();
        xv[2] = robotPose.phi();

        MRPT_END
}

/** This virtual function musts calculate the Jacobian F of the prediction
 * model.
 */
void CRangeBearingKFSLAM2D::OnTransitionJacobian(KFMatrix_VxV& F) const
{
        MRPT_START

        // The jacobian of the transition function:  dfv_dxv
        CActionRobotMovement2D::Ptr act2D = m_action->getBestMovementEstimation();
        CActionRobotMovement3D::Ptr act3D =
                m_action->getActionByClass<CActionRobotMovement3D>();

        if (act3D && act2D) THROW_EXCEPTION("Both 2D & 3D odometry are present!?!?")

        TPoint2D Ap;

        if (act3D)
        {
                Ap = TPoint2D(CPose2D(act3D->poseChange.mean));
        }
        else if (act2D)
        {
                Ap = TPoint2D(act2D->poseChange->getMeanVal());
        }
        else
        {
                // No odometry at all:
                F.setIdentity();  // Unit diagonal
                return;
        }

        const kftype cy = cos(m_xkk[2]);
        const kftype sy = sin(m_xkk[2]);

        const kftype Ax = Ap.x;
        const kftype Ay = Ap.y;

        F.setIdentity();  // Unit diagonal

        F.get_unsafe(0, 2) = -Ax * sy - Ay * cy;
        F.get_unsafe(1, 2) = Ax * cy - Ay * sy;

        MRPT_END
}

/** This virtual function musts calculate de noise matrix of the prediction
 * model.
 */
void CRangeBearingKFSLAM2D::OnTransitionNoise(KFMatrix_VxV& Q) const
{
        MRPT_START

        // The uncertainty of the 2D odometry, projected from the current position:
        CActionRobotMovement2D::Ptr act2D = m_action->getBestMovementEstimation();
        CActionRobotMovement3D::Ptr act3D =
                m_action->getActionByClass<CActionRobotMovement3D>();

        if (act3D && act2D) THROW_EXCEPTION("Both 2D & 3D odometry are present!?!?")

        CPosePDFGaussian odoIncr;

        if (!act3D && !act2D)
        {
                // Use constant Q:
                Q.zeros();
                ASSERT_(size_t(options.stds_Q_no_odo.size()) == Q.getColCount())

                for (size_t i = 0; i < 3; i++)
                        Q.get_unsafe(i, i) = square(options.stds_Q_no_odo[i]);
                return;
        }
        else
        {
                if (act2D)
                {
                        // 2D odometry:
                        odoIncr = CPosePDFGaussian(*act2D->poseChange);
                }
                else
                {
                        // 3D odometry:
                        odoIncr = CPosePDFGaussian(act3D->poseChange);
                }
        }

        odoIncr.rotateCov(m_xkk[2]);

        Q = odoIncr.cov;

        MRPT_END
}

void CRangeBearingKFSLAM2D::OnObservationModel(
        const vector_size_t& idx_landmarks_to_predict,
        vector_KFArray_OBS& out_predictions) const
{
        MRPT_START

        // Get the sensor pose relative to the robot:
        CObservationBearingRange::Ptr obs =
                m_SF->getObservationByClass<CObservationBearingRange>();
        ASSERTMSG_(
                obs,
                "*ERROR*: This method requires an observation of type "
                "CObservationBearingRange")
        const CPose2D sensorPoseOnRobot = CPose2D(obs->sensorLocationOnRobot);

        /* -------------------------------------------
           Equations, obtained using matlab, of the relative 2D position of a
          landmark (xi,yi), relative
                  to a robot 2D pose (x0,y0,phi)
                Refer to technical report "6D EKF derivation...", 2008

                x0 y0 phi0         % Robot's 2D pose
                x0s y0s phis      % Sensor's 2D pose relative to robot
                xi yi             % Absolute 2D landmark coordinates:

                Hx : dh_dxv   -> Jacobian of the observation model wrt the robot pose
                Hy : dh_dyi   -> Jacobian of the observation model wrt each landmark
          mean position

                Sizes:
                 h:  1x2
                 Hx: 2x3
                 Hy: 2x2
          ------------------------------------------- */
        // Mean of the prior of the robot pose:
        const CPose2D robotPose(m_xkk[0], m_xkk[1], m_xkk[2]);

        const size_t vehicle_size = get_vehicle_size();
        const size_t feature_size = get_feature_size();

        const CPose2D sensorPoseAbs = robotPose + sensorPoseOnRobot;

        // ---------------------------------------------------
        // Observation prediction
        // ---------------------------------------------------
        const size_t N = idx_landmarks_to_predict.size();
        out_predictions.resize(N);
        for (size_t i = 0; i < N; i++)
        {
                const size_t idx_lm = idx_landmarks_to_predict[i];
                ASSERTDEB_(idx_lm < this->getNumberOfLandmarksInTheMap());

                // Landmark absolute position in the map:
                const kftype xi = m_xkk[vehicle_size + feature_size * idx_lm + 0];
                const kftype yi = m_xkk[vehicle_size + feature_size * idx_lm + 1];

                const double Axi = xi - sensorPoseAbs.x();
                const double Ayi = yi - sensorPoseAbs.y();

                out_predictions[i][0] = sqrt(square(Axi) + square(Ayi));  // Range
                out_predictions[i][1] =
                        mrpt::math::wrapToPi(atan2(Ayi, Axi) - sensorPoseAbs.phi());  // Yaw
        }

        MRPT_END
}

void CRangeBearingKFSLAM2D::OnObservationJacobians(
        const size_t& idx_landmark_to_predict, KFMatrix_OxV& Hx,
        KFMatrix_OxF& Hy) const
{
        MRPT_START

        // Get the sensor pose relative to the robot:
        CObservationBearingRange::Ptr obs =
                m_SF->getObservationByClass<CObservationBearingRange>();
        ASSERTMSG_(
                obs,
                "*ERROR*: This method requires an observation of type "
                "CObservationBearingRange")
        const CPose2D sensorPoseOnRobot = CPose2D(obs->sensorLocationOnRobot);

        /* -------------------------------------------
           Equations, obtained using matlab, of the relative 2D position of a
          landmark (xi,yi), relative
                  to a robot 2D pose (x0,y0,phi)
                Refer to technical report "6D EKF derivation...", 2008

                x0 y0 phi0         % Robot's 2D pose
                x0s y0s phis      % Sensor's 2D pose relative to robot
                xi yi             % Absolute 2D landmark coordinates:

                Hx : dh_dxv   -> Jacobian of the observation model wrt the robot pose
                Hy : dh_dyi   -> Jacobian of the observation model wrt each landmark
          mean position

                Sizes:
                 h:  1x2
                 Hx: 2x3
                 Hy: 2x2
          ------------------------------------------- */
        // Mean of the prior of the robot pose:
        const CPose2D robotPose(m_xkk[0], m_xkk[1], m_xkk[2]);

        const size_t vehicle_size = get_vehicle_size();
        const size_t feature_size = get_feature_size();

        // Robot 6D pose:
        const kftype x0 = m_xkk[0];
        const kftype y0 = m_xkk[1];
        const kftype phi0 = m_xkk[2];

        const kftype cphi0 = cos(phi0);
        const kftype sphi0 = sin(phi0);

        // Sensor 2D pose on robot:
        const kftype x0s = sensorPoseOnRobot.x();
        const kftype y0s = sensorPoseOnRobot.y();
        const kftype phis = sensorPoseOnRobot.phi();

        const kftype cphi0s = cos(phi0 + phis);
        const kftype sphi0s = sin(phi0 + phis);

        const CPose2D sensorPoseAbs = robotPose + sensorPoseOnRobot;

        // Landmark absolute position in the map:
        const kftype xi =
                m_xkk[vehicle_size + feature_size * idx_landmark_to_predict + 0];
        const kftype yi =
                m_xkk[vehicle_size + feature_size * idx_landmark_to_predict + 1];

        // ---------------------------------------------------
        // Generate dhi_dxv: A 2x3 block
        // ---------------------------------------------------
        const kftype EXP1 =
                -2 * yi * y0s * cphi0 - 2 * yi * y0 + 2 * xi * y0s * sphi0 -
                2 * xi * x0 - 2 * xi * x0s * cphi0 - 2 * yi * x0s * sphi0 +
                2 * y0s * y0 * cphi0 - 2 * y0s * x0 * sphi0 + 2 * y0 * x0s * sphi0 +
                square(x0) + 2 * x0s * x0 * cphi0 + square(x0s) + square(y0s) +
                square(xi) + square(yi) + square(y0);
        const kftype sqrtEXP1_1 = kftype(1) / sqrt(EXP1);

        const kftype EXP2 = cphi0s * xi + sphi0s * yi - sin(phis) * y0s -
                                                y0 * sphi0s - x0s * cos(phis) - x0 * cphi0s;
        const kftype EXP2sq = square(EXP2);

        const kftype EXP3 = -sphi0s * xi + cphi0s * yi - cos(phis) * y0s -
                                                y0 * cphi0s + x0s * sin(phis) + x0 * sphi0s;
        const kftype EXP3sq = square(EXP3);

        const kftype EXP4 = kftype(1) / (1 + EXP3sq / EXP2sq);

        Hx.get_unsafe(0, 0) = (-xi - sphi0 * y0s + cphi0 * x0s + x0) * sqrtEXP1_1;
        Hx.get_unsafe(0, 1) = (-yi + cphi0 * y0s + y0 + sphi0 * x0s) * sqrtEXP1_1;
        Hx.get_unsafe(0, 2) =
                (y0s * xi * cphi0 + y0s * yi * sphi0 - y0 * y0s * sphi0 -
                 x0 * y0s * cphi0 + x0s * xi * sphi0 - x0s * yi * cphi0 +
                 y0 * x0s * cphi0 - x0s * x0 * sphi0) *
                sqrtEXP1_1;

        Hx.get_unsafe(1, 0) = (sphi0s / (EXP2) + (EXP3) / EXP2sq * cphi0s) * EXP4;
        Hx.get_unsafe(1, 1) = (-cphi0s / (EXP2) + (EXP3) / EXP2sq * sphi0s) * EXP4;
        Hx.get_unsafe(1, 2) =
                ((-cphi0s * xi - sphi0s * yi + y0 * sphi0s + x0 * cphi0s) / (EXP2) -
                 (EXP3) / EXP2sq *
                         (-sphi0s * xi + cphi0s * yi - y0 * cphi0s + x0 * sphi0s)) *
                EXP4;

        // ---------------------------------------------------
        // Generate dhi_dyi: A 2x2 block
        // ---------------------------------------------------
        Hy.get_unsafe(0, 0) = (xi + sphi0 * y0s - cphi0 * x0s - x0) * sqrtEXP1_1;
        Hy.get_unsafe(0, 1) = (yi - cphi0 * y0s - y0 - sphi0 * x0s) * sqrtEXP1_1;

        Hy.get_unsafe(1, 0) = (-sphi0s / (EXP2) - (EXP3) / EXP2sq * cphi0s) * EXP4;
        Hy.get_unsafe(1, 1) = (cphi0s / (EXP2) - (EXP3) / EXP2sq * sphi0s) * EXP4;

        MRPT_END
}

/** This is called between the KF prediction step and the update step, and the
 * application must return the observations and, when applicable, the data
 * association between these observations and the current map.
  *
  * \param out_z N vectors, each for one "observation" of length OBS_SIZE, N
 * being the number of "observations": how many observed landmarks for a map, or
 * just one if not applicable.
  * \param out_data_association An empty vector or, where applicable, a vector
 * where the i'th element corresponds to the position of the observation in the
 * i'th row of out_z within the system state vector (in the range
 * [0,getNumberOfLandmarksInTheMap()-1]), or -1 if it is a new map element and
 * we want to insert it at the end of this KF iteration.
  * \param in_S The full covariance matrix of the observation predictions (i.e.
 * the "innovation covariance matrix"). This is a M·O x M·O matrix with M=length
 * of "in_lm_indices_in_S".
  * \param in_lm_indices_in_S The indices of the map landmarks (range
 * [0,getNumberOfLandmarksInTheMap()-1]) that can be found in the matrix in_S.
  *
  *  This method will be called just once for each complete KF iteration.
  * \note It is assumed that the observations are independent, i.e. there are NO
 * cross-covariances between them.
  */
void CRangeBearingKFSLAM2D::OnGetObservationsAndDataAssociation(
        vector_KFArray_OBS& Z, vector_int& data_association,
        const vector_KFArray_OBS& all_predictions, const KFMatrix& S,
        const vector_size_t& lm_indices_in_S, const KFMatrix_OxO& R)
{
        MRPT_START

        // static const size_t vehicle_size = get_vehicle_size();
        static const size_t obs_size = get_observation_size();

        // Z: Observations
        CObservationBearingRange::TMeasurementList::const_iterator itObs;

        CObservationBearingRange::Ptr obs =
                m_SF->getObservationByClass<CObservationBearingRange>();
        ASSERTMSG_(
                obs,
                "*ERROR*: This method requires an observation of type "
                "CObservationBearingRange")

        const size_t N = obs->sensedData.size();
        Z.resize(N);

        size_t row;
        for (row = 0, itObs = obs->sensedData.begin();
                 itObs != obs->sensedData.end(); ++itObs, ++row)
        {
                // Fill one row in Z:
                Z[row][0] = itObs->range;
                Z[row][1] = itObs->yaw;
                ASSERTMSG_(
                        itObs->pitch == 0,
                        "ERROR: Observation contains pitch!=0 but this is 2D-SLAM!!!")
        }

        // Data association:
        // ---------------------
        data_association.assign(N, -1);  // Initially, all new landmarks

        // For each observed LM:
        vector_size_t obs_idxs_needing_data_assoc;
        obs_idxs_needing_data_assoc.reserve(N);

        {
                vector_int::iterator itDA;
                CObservationBearingRange::TMeasurementList::const_iterator itObs;
                size_t row;
                for (row = 0, itObs = obs->sensedData.begin(),
                        itDA = data_association.begin();
                         itObs != obs->sensedData.end(); ++itObs, ++itDA, ++row)
                {
                        // Fill data asociation: Using IDs!
                        if (itObs->landmarkID < 0)
                                obs_idxs_needing_data_assoc.push_back(row);
                        else
                        {
                                mrpt::utils::bimap<CLandmark::TLandmarkID,
                                                                   unsigned int>::iterator itID;
                                if ((itID = m_IDs.find_key(itObs->landmarkID)) != m_IDs.end())
                                        *itDA = itID->second;  // This row in Z corresponds to the
                                // i'th map element in the state
                                // vector:
                        }
                }
        }

        // ---- Perform data association ----
        //  Only for observation indices in "obs_idxs_needing_data_assoc"
        if (obs_idxs_needing_data_assoc.empty())
        {
                // We don't need to do DA:
                m_last_data_association = TDataAssocInfo();

                // Save them for the case the external user wants to access them:
                for (size_t idxObs = 0; idxObs < data_association.size(); idxObs++)
                {
                        int da = data_association[idxObs];
                        if (da >= 0)
                                m_last_data_association.results.associations[idxObs] =
                                        data_association[idxObs];
                }

#if STATS_EXPERIMENT  // DEBUG: Generate statistic info
                {
                        static CFileOutputStream fC("metric_stats_C.txt", true);
#if STATS_EXPERIMENT_ALSO_NC
                        static CFileOutputStream fNC("metric_stats_NC.txt", true);
#endif

                        vector_int::iterator itDA;
                        size_t idx_obs = 0;
                        for (itDA = data_association.begin();
                                 itDA != data_association.end(); ++itDA, ++idx_obs)
                        {
                                if (*itDA == -1) continue;
                                const size_t lm_idx_in_map = *itDA;
                                size_t valid_idx_pred =
                                        find_in_vector(lm_idx_in_map, lm_indices_in_S);

                                const KFArray_OBS& obs_mu = Z[idx_obs];

                                for (size_t idx_pred = 0; idx_pred < lm_indices_in_S.size();
                                         idx_pred++)
                                {
                                        const KFArray_OBS& lm_mu =
                                                all_predictions[lm_indices_in_S[idx_pred]];

                                        const size_t base_idx_in_S = obs_size * idx_pred;
                                        KFMatrix_OxO lm_cov;
                                        S.extractMatrix(base_idx_in_S, base_idx_in_S, lm_cov);

                                        double md, log_pdf;
                                        mahalanobisDistance2AndLogPDF(
                                                lm_mu - obs_mu, lm_cov, md, log_pdf);

                                        if (valid_idx_pred == idx_pred)
                                                fC.printf("%e %e\n", md, log_pdf);
                                        else
                                        {
#if STATS_EXPERIMENT_ALSO_NC
                                                fNC.printf("%e %e\n", md, log_pdf);
#endif
                                        }
                                }
                        }
                }
#endif
        }
        else
        {
                // Build a Z matrix with the observations that need dat.assoc:
                const size_t nObsDA = obs_idxs_needing_data_assoc.size();

                CMatrixTemplateNumeric<kftype> Z_obs_means(nObsDA, obs_size);
                for (size_t i = 0; i < nObsDA; i++)
                {
                        const size_t idx = obs_idxs_needing_data_assoc[i];
                        for (unsigned k = 0; k < obs_size; k++)
                                Z_obs_means.get_unsafe(i, k) = Z[idx][k];
                }

                // Vehicle uncertainty
                KFMatrix_VxV Pxx(UNINITIALIZED_MATRIX);
                m_pkk.extractMatrix(0, 0, Pxx);

                // Build predictions:
                // ---------------------------
                const size_t nPredictions = lm_indices_in_S.size();
                m_last_data_association.clear();

                // S is the covariance of the predictions:
                m_last_data_association.Y_pred_covs = S;

                // The means:
                m_last_data_association.Y_pred_means.setSize(nPredictions, obs_size);
                for (size_t q = 0; q < nPredictions; q++)
                {
                        const size_t i = lm_indices_in_S[q];
                        for (size_t w = 0; w < obs_size; w++)
                                m_last_data_association.Y_pred_means.get_unsafe(q, w) =
                                        all_predictions[i][w];
                        m_last_data_association.predictions_IDs.push_back(
                                i);  // for the conversion of indices...
                }

                // Do Dat. Assoc :
                // ---------------------------
                if (nPredictions)
                {
                        CMatrixDouble Z_obs_cov = CMatrixDouble(R);

                        mrpt::slam::data_association_full_covariance(
                                Z_obs_means,  // Z_obs_cov,
                                m_last_data_association.Y_pred_means,
                                m_last_data_association.Y_pred_covs,
                                m_last_data_association.results, options.data_assoc_method,
                                options.data_assoc_metric, options.data_assoc_IC_chi2_thres,
                                true,  // Use KD-tree
                                m_last_data_association.predictions_IDs,
                                options.data_assoc_IC_metric,
                                options.data_assoc_IC_ml_threshold);

                        // Return pairings to the main KF algorithm:
                        for (map<size_t, size_t>::const_iterator it =
                                         m_last_data_association.results.associations.begin();
                                 it != m_last_data_association.results.associations.end(); ++it)
                                data_association[it->first] = it->second;
                }
        }
        // ---- End of data association ----

        MRPT_END
}

/** This virtual function musts normalize the state vector and covariance matrix
 * (only if its necessary).
 */
void CRangeBearingKFSLAM2D::OnNormalizeStateVector()
{
        // Check angles:
        mrpt::math::wrapToPiInPlace(m_xkk[2]);
}

/*---------------------------------------------------------------
                                                loadOptions
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::loadOptions(const mrpt::utils::CConfigFileBase& ini)
{
        // Main
        options.loadFromConfigFile(ini, "RangeBearingKFSLAM");
        KF_options.loadFromConfigFile(ini, "RangeBearingKFSLAM_KalmanFilter");
}

/*---------------------------------------------------------------
                                                loadOptions
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::TOptions::loadFromConfigFile(
        const mrpt::utils::CConfigFileBase& source, const std::string& section)
{
        stds_Q_no_odo[2] = RAD2DEG(stds_Q_no_odo[2]);

        source.read_vector(section, "stds_Q_no_odo", stds_Q_no_odo, stds_Q_no_odo);
        ASSERT_(stds_Q_no_odo.size() == 3)

        stds_Q_no_odo[2] = DEG2RAD(stds_Q_no_odo[2]);

        std_sensor_range =
                source.read_float(section, "std_sensor_range", std_sensor_range);
        std_sensor_yaw = DEG2RAD(
                source.read_float(
                        section, "std_sensor_yaw_deg", RAD2DEG(std_sensor_yaw)));

        MRPT_LOAD_CONFIG_VAR(quantiles_3D_representation, float, source, section);
        MRPT_LOAD_CONFIG_VAR(create_simplemap, bool, source, section);

        data_assoc_method = source.read_enum<TDataAssociationMethod>(
                section, "data_assoc_method", data_assoc_method);
        data_assoc_metric = source.read_enum<TDataAssociationMetric>(
                section, "data_assoc_metric", data_assoc_metric);
        data_assoc_IC_metric = source.read_enum<TDataAssociationMetric>(
                section, "data_assoc_IC_metric", data_assoc_IC_metric);

        MRPT_LOAD_CONFIG_VAR(data_assoc_IC_chi2_thres, double, source, section);
        MRPT_LOAD_CONFIG_VAR(data_assoc_IC_ml_threshold, double, source, section);
}

/*---------------------------------------------------------------
                                                Constructor
  ---------------------------------------------------------------*/
CRangeBearingKFSLAM2D::TOptions::TOptions()
        : stds_Q_no_odo(3, 0),
          std_sensor_range(0.1f),
          std_sensor_yaw(DEG2RAD(0.5f)),
          quantiles_3D_representation(3),
          create_simplemap(false),
          data_assoc_method(assocNN),
          data_assoc_metric(metricMaha),
          data_assoc_IC_chi2_thres(0.99),
          data_assoc_IC_metric(metricMaha),
          data_assoc_IC_ml_threshold(0.0)

{
        stds_Q_no_odo[0] = 0.10f;
        stds_Q_no_odo[1] = 0.10f;
        stds_Q_no_odo[2] = DEG2RAD(4.0f);
}

/*---------------------------------------------------------------
                                                dumpToTextStream
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::TOptions::dumpToTextStream(CStream& out) const
{
        out.printf(
                "\n----------- [CRangeBearingKFSLAM2D::TOptions] ------------ \n\n");

        out.printf(
                "data_assoc_method                       = %s\n",
                TEnumType<TDataAssociationMethod>::value2name(data_assoc_method)
                        .c_str());
        out.printf(
                "data_assoc_metric                       = %s\n",
                TEnumType<TDataAssociationMetric>::value2name(data_assoc_metric)
                        .c_str());
        out.printf(
                "data_assoc_IC_chi2_thres                = %.06f\n",
                data_assoc_IC_chi2_thres);
        out.printf(
                "data_assoc_IC_metric                    = %s\n",
                TEnumType<TDataAssociationMetric>::value2name(data_assoc_IC_metric)
                        .c_str());
        out.printf(
                "data_assoc_IC_ml_threshold              = %.06f\n",
                data_assoc_IC_ml_threshold);

        out.printf("\n");
}

void CRangeBearingKFSLAM2D::OnInverseObservationModel(
        const KFArray_OBS& in_z, KFArray_FEAT& yn, KFMatrix_FxV& dyn_dxv,
        KFMatrix_FxO& dyn_dhn) const
{
        MRPT_START

        /* -------------------------------------------
           Equations, obtained using matlab

                x0 y0 phi0      % Robot's 2D pose
                x0s y0s phis    % Sensor's 2D pose relative to robot
                hr ha           % Observation hn: range, yaw

                xi yi           % Absolute 2D landmark coordinates:

                dyn_dxv   -> Jacobian of the inv. observation model wrt the robot pose
                dyn_dhn   -> Jacobian of the inv. observation model wrt each landmark
          observation

                Sizes:
                 hn:      1x2  <--
                 yn:      1x2  -->
                 dyn_dxv: 2x3  -->
                 dyn_dhn: 2x2  -->
          ------------------------------------------- */

        CObservationBearingRange::Ptr obs =
                m_SF->getObservationByClass<CObservationBearingRange>();
        ASSERTMSG_(
                obs,
                "*ERROR*: This method requires an observation of type "
                "CObservationBearingRange")
        const CPose2D sensorPoseOnRobot = CPose2D(obs->sensorLocationOnRobot);

        // Mean of the prior of the robot pose:
        const TPose2D robotPose(m_xkk[0], m_xkk[1], m_xkk[2]);

        // Robot 2D pose:
        const kftype x0 = m_xkk[0];
        const kftype y0 = m_xkk[1];
        const kftype phi0 = m_xkk[2];

        const kftype cphi0 = cos(phi0);
        const kftype sphi0 = sin(phi0);

        // Sensor 6D pose on robot:
        const kftype x0s = sensorPoseOnRobot.x();
        const kftype y0s = sensorPoseOnRobot.y();
        const kftype phis = sensorPoseOnRobot.phi();

        const kftype hr = in_z[0];
        const kftype ha = in_z[1];

        const kftype cphi_0sa = cos(phi0 + phis + ha);
        const kftype sphi_0sa = sin(phi0 + phis + ha);

        // Compute the mean 2D absolute coordinates:
        yn[0] = hr * cphi_0sa + cphi0 * x0s - sphi0 * y0s + x0;
        yn[1] = hr * sphi_0sa + sphi0 * x0s + cphi0 * y0s + y0;

        // Jacobian wrt xv:
        dyn_dxv.get_unsafe(0, 0) = 1;
        dyn_dxv.get_unsafe(0, 1) = 0;
        dyn_dxv.get_unsafe(0, 2) = -hr * sphi_0sa - sphi0 * x0s - cphi0 * y0s;

        dyn_dxv.get_unsafe(1, 0) = 0;
        dyn_dxv.get_unsafe(1, 1) = 1;
        dyn_dxv.get_unsafe(1, 2) = hr * cphi_0sa + cphi0 * x0s - sphi0 * y0s;

        // Jacobian wrt hn:
        dyn_dhn.get_unsafe(0, 0) = cphi_0sa;
        dyn_dhn.get_unsafe(0, 1) = -hr * sphi_0sa;

        dyn_dhn.get_unsafe(1, 0) = sphi_0sa;
        dyn_dhn.get_unsafe(1, 1) = hr * cphi_0sa;

        MRPT_END
}

/** If applicable to the given problem, do here any special handling of adding a
 * new landmark to the map.
  * \param in_obsIndex The index of the observation whose inverse sensor is to
 * be computed. It corresponds to the row in in_z where the observation can be
 * found.
  * \param in_idxNewFeat The index that this new feature will have in the state
 * vector (0:just after the vehicle state, 1: after that,...). Save this number
 * so data association can be done according to these indices.
  * \sa OnInverseObservationModel
  */
void CRangeBearingKFSLAM2D::OnNewLandmarkAddedToMap(
        const size_t in_obsIdx, const size_t in_idxNewFeat)
{
        MRPT_START

        CObservationBearingRange::Ptr obs =
                m_SF->getObservationByClass<CObservationBearingRange>();
        ASSERTMSG_(
                obs,
                "*ERROR*: This method requires an observation of type "
                "CObservationBearingRange")

        // ----------------------------------------------
        // introduce in the lists of ID<->index in map:
        // ----------------------------------------------
        ASSERT_(in_obsIdx < obs->sensedData.size())

        if (obs->sensedData[in_obsIdx].landmarkID >= 0)
        {
                // The sensor provides us a LM ID... use it:
                m_IDs.insert(obs->sensedData[in_obsIdx].landmarkID, in_idxNewFeat);
        }
        else
        {
                // Features do not have IDs... use indices:
                m_IDs.insert(in_idxNewFeat, in_idxNewFeat);
        }

        m_last_data_association.newly_inserted_landmarks[in_obsIdx] =
                in_idxNewFeat;  // Just for stats, etc...

        MRPT_END
}

/*---------------------------------------------------------------
                                                getAs3DObject
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::getAs3DObject(
        mrpt::opengl::CSetOfObjects::Ptr& outObj) const
{
        outObj->clear();

        // ------------------------------------------------
        //  Add the XYZ corner for the current area:
        // ------------------------------------------------
        outObj->insert(opengl::stock_objects::CornerXYZ());

        // 2D ellipsoid for robot pose:
        CPoint2DPDFGaussian pointGauss;
        pointGauss.mean.x(m_xkk[0]);
        pointGauss.mean.y(m_xkk[1]);
        CMatrixTemplateNumeric<kftype> COV;
        m_pkk.extractMatrix(0, 0, 2, 2, COV);
        pointGauss.cov = COV;

        {
                opengl::CEllipsoid::Ptr ellip =
                        mrpt::make_aligned_shared<opengl::CEllipsoid>();

                ellip->setPose(pointGauss.mean);
                ellip->setCovMatrix(pointGauss.cov);
                ellip->enableDrawSolid3D(false);
                ellip->setQuantiles(options.quantiles_3D_representation);
                ellip->set3DsegmentsCount(10);
                ellip->setColor(1, 0, 0);

                outObj->insert(ellip);
        }

        // 2D ellipsoids for landmarks:
        const size_t nLMs = (m_xkk.size() - 3) / 2;
        for (size_t i = 0; i < nLMs; i++)
        {
                pointGauss.mean.x(m_xkk[3 + 2 * i + 0]);
                pointGauss.mean.y(m_xkk[3 + 2 * i + 1]);
                m_pkk.extractMatrix(3 + 2 * i, 3 + 2 * i, 2, 2, COV);
                pointGauss.cov = COV;

                opengl::CEllipsoid::Ptr ellip =
                        mrpt::make_aligned_shared<opengl::CEllipsoid>();

                ellip->setName(format("%u", static_cast<unsigned int>(i)));
                ellip->enableShowName(true);
                ellip->setPose(pointGauss.mean);
                ellip->setCovMatrix(pointGauss.cov);
                ellip->enableDrawSolid3D(false);
                ellip->setQuantiles(options.quantiles_3D_representation);
                ellip->set3DsegmentsCount(10);

                ellip->setColor(0, 0, 1);

                outObj->insert(ellip);
        }
}

/*---------------------------------------------------------------
                          saveMapAndPathRepresentationAsMATLABFile
  ---------------------------------------------------------------*/
void CRangeBearingKFSLAM2D::saveMapAndPath2DRepresentationAsMATLABFile(
        const string& fil, float stdCount, const string& styleLandmarks,
        const string& stylePath, const string& styleRobot) const
{
        FILE* f = os::fopen(fil.c_str(), "wt");
        if (!f) return;

        CMatrixDouble cov(2, 2);
        CVectorDouble mean(2);

        // Header:
        os::fprintf(
                f,
                "%%--------------------------------------------------------------------"
                "\n");
        os::fprintf(f, "%% File automatically generated using the MRPT method:\n");
        os::fprintf(
                f,
                "%% "
                "'CRangeBearingKFSLAM2D::saveMapAndPath2DRepresentationAsMATLABFile'"
                "\n");
        os::fprintf(f, "%%\n");
        os::fprintf(f, "%%                        ~ MRPT ~\n");
        os::fprintf(
                f, "%%  Jose Luis Blanco Claraco, University of Malaga @ 2008\n");
        os::fprintf(f, "%%      http://www.mrpt.org/     \n");
        os::fprintf(
                f,
                "%%--------------------------------------------------------------------"
                "\n");

        // Main code:
        os::fprintf(f, "hold on;\n\n");

        size_t i, nLMs = (m_xkk.size() - get_vehicle_size()) / get_feature_size();

        for (i = 0; i < nLMs; i++)
        {
                size_t idx = get_vehicle_size() + i * get_feature_size();

                cov(0, 0) = m_pkk(idx + 0, idx + 0);
                cov(1, 1) = m_pkk(idx + 1, idx + 1);
                cov(0, 1) = cov(1, 0) = m_pkk(idx + 0, idx + 1);

                mean[0] = m_xkk[idx + 0];
                mean[1] = m_xkk[idx + 1];

                // Command to draw the 2D ellipse:
                os::fprintf(
                        f, "%s",
                        math::MATLAB_plotCovariance2D(cov, mean, stdCount, styleLandmarks)
                                .c_str());
        }

        // Now: the robot path:
        // ------------------------------
        if (m_SFs.size())
        {
                os::fprintf(f, "\nROB_PATH=[");
                for (i = 0; i < m_SFs.size(); i++)
                {
                        CSensoryFrame::Ptr dummySF;
                        CPose3DPDF::Ptr pdf3D;
                        m_SFs.get(i, pdf3D, dummySF);

                        CPose3D p;
                        pdf3D->getMean(p);
                        CPoint3D pnt3D(p);  // 6D -> 3D only

                        os::fprintf(f, "%.04f %.04f", pnt3D.x(), pnt3D.y());
                        if (i < (m_SFs.size() - 1)) os::fprintf(f, ";");
                }
                os::fprintf(f, "];\n");

                os::fprintf(
                        f, "plot(ROB_PATH(:,1),ROB_PATH(:,2),'%s');\n", stylePath.c_str());
        }

        // The robot pose:
        cov(0, 0) = m_pkk(0, 0);
        cov(1, 1) = m_pkk(1, 1);
        cov(0, 1) = cov(1, 0) = m_pkk(0, 1);

        mean[0] = m_xkk[0];
        mean[1] = m_xkk[1];

        os::fprintf(
                f, "%s",
                math::MATLAB_plotCovariance2D(cov, mean, stdCount, styleRobot).c_str());

        os::fprintf(f, "\naxis equal;\n");
        os::fclose(f);
}

/** Computes A=A-B, which may need to be re-implemented depending on the
 * topology of the individual scalar components (eg, angles).
  */
void CRangeBearingKFSLAM2D::OnSubstractObservationVectors(
        KFArray_OBS& A, const KFArray_OBS& B) const
{
        A[0] -= B[0];
        A[1] -= B[1];
        mrpt::math::wrapToPiInPlace(A[1]);
}

/** Return the observation NOISE covariance matrix, that is, the model of the
 * Gaussian additive noise of the sensor.
  * \param out_R The noise covariance matrix. It might be non diagonal, but
 * it'll usually be.
  */
void CRangeBearingKFSLAM2D::OnGetObservationNoise(KFMatrix_OxO& out_R) const
{
        out_R(0, 0) = square(options.std_sensor_range);
        out_R(1, 1) = square(options.std_sensor_yaw);
}

/** This will be called before OnGetObservationsAndDataAssociation to allow the
 * application to reduce the number of covariance landmark predictions to be
 * made.
  *  For example, features which are known to be "out of sight" shouldn't be
 * added to the output list to speed up the calculations.
  * \param in_all_prediction_means The mean of each landmark predictions; the
 * computation or not of the corresponding covariances is what we're trying to
 * determined with this method.
  * \param out_LM_indices_to_predict The list of landmark indices in the map
 * [0,getNumberOfLandmarksInTheMap()-1] that should be predicted.
  * \note This is not a pure virtual method, so it should be implemented only if
 * desired. The default implementation returns a vector with all the landmarks
 * in the map.
  * \sa OnGetObservations, OnDataAssociation
  */
void CRangeBearingKFSLAM2D::OnPreComputingPredictions(
        const vector_KFArray_OBS& prediction_means,
        vector_size_t& out_LM_indices_to_predict) const
{
        CObservationBearingRange::Ptr obs =
                m_SF->getObservationByClass<CObservationBearingRange>();
        ASSERTMSG_(
                obs,
                "*ERROR*: This method requires an observation of type "
                "CObservationBearingRange")

        const double sensor_max_range = obs->maxSensorDistance;
        const double fov_yaw = obs->fieldOfView_yaw;

        const double max_vehicle_loc_uncertainty =
                4 * std::sqrt(m_pkk.get_unsafe(0, 0) + m_pkk.get_unsafe(1, 1));
        const double max_vehicle_ang_uncertainty =
                4 * std::sqrt(m_pkk.get_unsafe(2, 2));

        out_LM_indices_to_predict.clear();
        for (size_t i = 0; i < prediction_means.size(); i++)
#if (!STATS_EXPERIMENT)  // In the experiment we force errors too far and some
                // predictions are out of range, so just generate all
                // of them:
                if (prediction_means[i][0] <
                                (1.5 + sensor_max_range + max_vehicle_loc_uncertainty +
                                 4 * options.std_sensor_range) &&
                        fabs(prediction_means[i][1]) <
                                (DEG2RAD(20) + 0.5 * fov_yaw + max_vehicle_ang_uncertainty +
                                 4 * options.std_sensor_yaw))
#endif
                {
                        out_LM_indices_to_predict.push_back(i);
                }
}

/** Only called if using a numeric approximation of the transition Jacobian,
 * this method must return the increments in each dimension of the vehicle state
 * vector while estimating the Jacobian.
  */
void CRangeBearingKFSLAM2D::OnTransitionJacobianNumericGetIncrements(
        KFArray_VEH& out_increments) const
{
        for (size_t i = 0; i < get_vehicle_size(); i++) out_increments[i] = 1e-6;
}

/** Only called if using a numeric approximation of the observation Jacobians,
 * this method must return the increments in each dimension of the vehicle state
 * vector while estimating the Jacobian.
  */
void CRangeBearingKFSLAM2D::OnObservationJacobiansNumericGetIncrements(
        KFArray_VEH& out_veh_increments, KFArray_FEAT& out_feat_increments) const
{
        for (size_t i = 0; i < get_vehicle_size(); i++)
                out_veh_increments[i] = 1e-6;
        for (size_t i = 0; i < get_feature_size(); i++)
                out_feat_increments[i] = 1e-6;
}