/*
    * Copyright (C) 2021 Alberto Irurueta Carro (alberto@irurueta.com)
    *
    * Licensed under the Apache License, Version 2.0 (the "License");
    * you may not use this file except in compliance with the License.
    * You may obtain a copy of the License at
    *
    *         http://www.apache.org/licenses/LICENSE-2.0
    *
   * Unless required by applicable law or agreed to in writing, software
   * distributed under the License is distributed on an "AS IS" BASIS,
   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
   * See the License for the specific language governing permissions and
   * limitations under the License.
   */
  package com.irurueta.navigation.inertial.calibration.intervals.thresholdfactor;
  
  import com.irurueta.algebra.Matrix;
  import com.irurueta.navigation.LockedException;
  import com.irurueta.navigation.NotReadyException;
  import com.irurueta.navigation.inertial.INSLooselyCoupledKalmanInitializerConfig;
  import com.irurueta.navigation.inertial.INSTightlyCoupledKalmanInitializerConfig;
  import com.irurueta.navigation.inertial.calibration.AccelerometerNoiseRootPsdSource;
  import com.irurueta.navigation.inertial.calibration.BodyKinematicsSequence;
  import com.irurueta.navigation.inertial.calibration.CalibrationException;
  import com.irurueta.navigation.inertial.calibration.GyroscopeNoiseRootPsdSource;
  import com.irurueta.navigation.inertial.calibration.StandardDeviationBodyKinematics;
  import com.irurueta.navigation.inertial.calibration.StandardDeviationTimedBodyKinematics;
  import com.irurueta.navigation.inertial.calibration.TimedBodyKinematics;
  import com.irurueta.navigation.inertial.calibration.accelerometer.AccelerometerCalibratorMeasurementType;
  import com.irurueta.navigation.inertial.calibration.accelerometer.AccelerometerNonLinearCalibrator;
  import com.irurueta.navigation.inertial.calibration.accelerometer.KnownBiasAccelerometerCalibrator;
  import com.irurueta.navigation.inertial.calibration.accelerometer.OrderedStandardDeviationBodyKinematicsAccelerometerCalibrator;
  import com.irurueta.navigation.inertial.calibration.accelerometer.QualityScoredAccelerometerCalibrator;
  import com.irurueta.navigation.inertial.calibration.accelerometer.UnknownBiasAccelerometerCalibrator;
  import com.irurueta.navigation.inertial.calibration.accelerometer.UnorderedStandardDeviationBodyKinematicsAccelerometerCalibrator;
  import com.irurueta.navigation.inertial.calibration.generators.AccelerometerAndGyroscopeMeasurementsGenerator;
  import com.irurueta.navigation.inertial.calibration.generators.AccelerometerAndGyroscopeMeasurementsGeneratorListener;
  import com.irurueta.navigation.inertial.calibration.gyroscope.AccelerometerDependentGyroscopeCalibrator;
  import com.irurueta.navigation.inertial.calibration.gyroscope.GyroscopeCalibratorMeasurementOrSequenceType;
  import com.irurueta.navigation.inertial.calibration.gyroscope.GyroscopeNonLinearCalibrator;
  import com.irurueta.navigation.inertial.calibration.gyroscope.OrderedBodyKinematicsSequenceGyroscopeCalibrator;
  import com.irurueta.navigation.inertial.calibration.gyroscope.QualityScoredGyroscopeCalibrator;
  import com.irurueta.navigation.inertial.calibration.gyroscope.UnknownBiasGyroscopeCalibrator;
  import com.irurueta.navigation.inertial.calibration.intervals.TriadStaticIntervalDetector;
  import com.irurueta.units.Acceleration;
  import com.irurueta.units.AccelerationUnit;
  import com.irurueta.units.Time;
  
  import java.util.ArrayList;
  import java.util.List;
  
  /**
   * Optimizes the threshold factor for interval detection of accelerometer and gyroscope
   * data based on results of calibration.
   * Implementations of this class will attempt to find the best threshold factor
   * between the provided range of values.
   * Only accelerometer calibrators based on unknown orientation are supported (in other terms,
   * calibrators must be {@link AccelerometerNonLinearCalibrator} and must support
   * {@link AccelerometerCalibratorMeasurementType#STANDARD_DEVIATION_BODY_KINEMATICS}).
   * Only gyroscope calibrators based on unknown orientation are supported (in other terms,
   * calibrators must be {@link GyroscopeNonLinearCalibrator} and must support
   * {@link GyroscopeCalibratorMeasurementOrSequenceType#BODY_KINEMATICS_SEQUENCE}).
   */
  public abstract class AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizer extends
          IntervalDetectorThresholdFactorOptimizer<TimedBodyKinematics,
                  AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizerDataSource> implements
          AccelerometerNoiseRootPsdSource, GyroscopeNoiseRootPsdSource {
  
      /**
       * Default minimum threshold factor.
       */
      public static final double DEFAULT_MIN_THRESHOLD_FACTOR = 2.0;
  
      /**
       * Default maximum threshold factor.
       */
      public static final double DEFAULT_MAX_THRESHOLD_FACTOR = 10.0;
  
      /**
       * Minimum threshold factor.
       */
      protected double minThresholdFactor = DEFAULT_MIN_THRESHOLD_FACTOR;
  
      /**
       * Maximum threshold factor.
       */
      protected double maxThresholdFactor = DEFAULT_MAX_THRESHOLD_FACTOR;
  
      /**
       * Accelerometer calibrator.
       */
      private AccelerometerNonLinearCalibrator accelerometerCalibrator;
  
      /**
       * Gyroscope calibrator.
       */
      private GyroscopeNonLinearCalibrator gyroscopeCalibrator;
  
     /**
      * A measurement generator for accelerometer and gyroscope calibrators.
      */
     private AccelerometerAndGyroscopeMeasurementsGenerator generator;
 
     /**
      * Generated measurements to be used for accelerometer calibration.
      */
     private List<StandardDeviationBodyKinematics> measurements;
 
     /**
      * Generated sequences to be used for gyroscope calibration.
      */
     private List<BodyKinematicsSequence<StandardDeviationTimedBodyKinematics>> sequences;
 
     /**
      * Mapper to convert {@link StandardDeviationBodyKinematics} measurements into
      * quality scores.
      */
     private QualityScoreMapper<StandardDeviationBodyKinematics> accelerometerQualityScoreMapper =
             new DefaultAccelerometerQualityScoreMapper();
 
     /**
      * Mapper to convert {@link BodyKinematicsSequence} sequences of {@link StandardDeviationTimedBodyKinematics}
      * into quality scores.
      */
     private QualityScoreMapper<BodyKinematicsSequence<StandardDeviationTimedBodyKinematics>>
             gyroscopeQualityScoreMapper = new DefaultGyroscopeQualityScoreMapper();
 
     /**
      * Rule to convert accelerometer and gyroscope MSE values into a single global
      * MSE value.
      */
     private AccelerometerAndGyroscopeMseRule mseRule =
             new DefaultAccelerometerAndGyroscopeMseRule();
 
     /**
      * Estimated norm of gyroscope noise root PSD (Power Spectral Density)
      * expressed as (rad * s^-0.5).
      */
     private double angularSpeedNoiseRootPsd;
 
     /**
      * Accelerometer base noise level that has been detected during
      * initialization of the best solution that has been found expressed in
      * meters per squared second (m/s^2).
      * This is equal to the standard deviation of the accelerometer measurements
      * during the initialization phase.
      */
     private double baseNoiseLevel;
 
     /**
      * Threshold to determine static/dynamic period changes expressed in
      * meters per squared second (m/s^2) for the best calibration solution that
      * has been found.
      */
     private double threshold;
 
     /**
      * Estimated covariance matrix for estimated accelerometer parameters.
      */
     private Matrix estimatedAccelerometerCovariance;
 
     /**
      * Estimated accelerometer scale factors and cross-coupling errors.
      * This is the product of matrix Ta containing cross-coupling errors and Ka
      * containing scaling factors.
      * So that:
      * <pre>
      *     Ma = [sx    mxy  mxz] = Ta*Ka
      *          [myx   sy   myz]
      *          [mzx   mzy  sz ]
      * </pre>
      * Where:
      * <pre>
      *     Ka = [sx 0   0 ]
      *          [0  sy  0 ]
      *          [0  0   sz]
      * </pre>
      * and
      * <pre>
      *     Ta = [1          -alphaXy    alphaXz ]
      *          [alphaYx    1           -alphaYz]
      *          [-alphaZx   alphaZy     1       ]
      * </pre>
      * Hence:
      * <pre>
      *     Ma = [sx    mxy  mxz] = Ta*Ka =  [sx             -sy * alphaXy   sz * alphaXz ]
      *          [myx   sy   myz]            [sx * alphaYx   sy              -sz * alphaYz]
      *          [mzx   mzy  sz ]            [-sx * alphaZx  sy * alphaZy    sz           ]
      * </pre>
      * This instance allows any 3x3 matrix. However, typically alphaYx, alphaZx and alphaZy
      * are considered to be zero if the accelerometer z-axis is assumed to be the same
      * as the body z-axis. When this is assumed, myx = mzx = mzy = 0 and the Ma matrix
      * becomes upper diagonal:
      * <pre>
      *     Ma = [sx    mxy  mxz]
      *          [0     sy   myz]
      *          [0     0    sz ]
      * </pre>
      * Values of this matrix are unit-less.
      */
     private Matrix estimatedAccelerometerMa;
 
     /**
      * Estimated accelerometer biases for each IMU axis expressed in meter per squared
      * second (m/s^2).
      */
     private double[] estimatedAccelerometerBiases;
 
     /**
      * Estimated covariance matrix for estimated gyroscope parameters.
      */
     private Matrix estimatedGyroscopeCovariance;
 
     /**
      * Estimated angular rate biases for each IMU axis expressed in radians per
      * second (rad/s).
      */
     private double[] estimatedGyroscopeBiases;
 
     /**
      * Estimated gyroscope scale factors and cross-coupling errors.
      * This is the product of matrix Tg containing cross-coupling errors and Kg
      * containing scaling factors.
      * So that:
      * <pre>
      *     Mg = [sx    mxy  mxz] = Tg*Kg
      *          [myx   sy   myz]
      *          [mzx   mzy  sz ]
      * </pre>
      * Where:
      * <pre>
      *     Kg = [sx 0   0 ]
      *          [0  sy  0 ]
      *          [0  0   sz]
      * </pre>
      * and
      * <pre>
      *     Tg = [1          -alphaXy    alphaXz ]
      *          [alphaYx    1           -alphaYz]
      *          [-alphaZx   alphaZy     1       ]
      * </pre>
      * Hence:
      * <pre>
      *     Mg = [sx    mxy  mxz] = Tg*Kg =  [sx             -sy * alphaXy   sz * alphaXz ]
      *          [myx   sy   myz]            [sx * alphaYx   sy              -sz * alphaYz]
      *          [mzx   mzy  sz ]            [-sx * alphaZx  sy * alphaZy    sz           ]
      * </pre>
      * This instance allows any 3x3 matrix. However, typically alphaYx, alphaZx and alphaZy
      * are considered to be zero if the gyroscope z-axis is assumed to be the same
      * as the body z-axis. When this is assumed, myx = mzx = mzy = 0 and the Mg matrix
      * becomes upper diagonal:
      * <pre>
      *     Mg = [sx    mxy  mxz]
      *          [0     sy   myz]
      *          [0     0    sz ]
      * </pre>
      * Values of this matrix are unit-less.
      */
     private Matrix estimatedGyroscopeMg;
 
     /**
      * Estimated G-dependent cross-biases introduced on the gyroscope by the
      * specific forces sensed by the accelerometer.
      * This instance allows any 3x3 matrix.
      */
     private Matrix estimatedGyroscopeGg;
 
     /**
      * Constructor.
      */
     protected AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizer() {
         initialize();
     }
 
     /**
      * Constructor.
      *
      * @param dataSource instance in charge of retrieving data for this optimizer.
      */
     protected AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizer(
             final AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizerDataSource dataSource) {
         super(dataSource);
         initialize();
     }
 
     /**
      * Constructor.
      *
      * @param accelerometerCalibrator an accelerometer calibrator to be used to
      *                                optimize its Mean Square Error (MSE).
      * @param gyroscopeCalibrator     a gyroscope calibrator to be used to optimize
      *                                its Mean Square Error (MSE).
      * @throws IllegalArgumentException if accelerometer calibrator does not use
      *                                  {@link StandardDeviationBodyKinematics} measurements
      *                                  or if gyroscope calibrator does not use
      *                                  {@link BodyKinematicsSequence} sequences of
      *                                  {@link StandardDeviationTimedBodyKinematics}.
      */
     protected AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizer(
             final AccelerometerNonLinearCalibrator accelerometerCalibrator,
             final GyroscopeNonLinearCalibrator gyroscopeCalibrator) {
         try {
             setAccelerometerCalibrator(accelerometerCalibrator);
             setGyroscopeCalibrator(gyroscopeCalibrator);
         } catch (final LockedException ignore) {
             // never happens
         }
         initialize();
     }
 
     /**
      * Constructor.
      *
      * @param dataSource              instance in charge of retrieving data for this
      *                                optimizer.
      * @param accelerometerCalibrator an accelerometer calibrator to be used to
      *                                optimize its Mean Square Error (MSE).
      * @param gyroscopeCalibrator     a gyroscope calibrator to be used to optimize
      *                                its Mean Square Error (MSE).
      * @throws IllegalArgumentException if accelerometer calibrator does not use
      *                                  {@link StandardDeviationBodyKinematics} measurements
      *                                  or if gyroscope calibrator does not use
      *                                  {@link BodyKinematicsSequence} sequences of
      *                                  {@link StandardDeviationTimedBodyKinematics}.
      */
     protected AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizer(
             final AccelerometerAndGyroscopeIntervalDetectorThresholdFactorOptimizerDataSource dataSource,
             final AccelerometerNonLinearCalibrator accelerometerCalibrator,
             final GyroscopeNonLinearCalibrator gyroscopeCalibrator) {
         super(dataSource);
         try {
             setAccelerometerCalibrator(accelerometerCalibrator);
             setGyroscopeCalibrator(gyroscopeCalibrator);
         } catch (final LockedException ignore) {
             // never happens
         }
         initialize();
     }
 
 
     /**
      * Gets provided accelerometer calibrator to be used to optimize its Mean Square Error (MSE).
      *
      * @return accelerometer calibrator to be used to optimize its MSE.
      */
     public AccelerometerNonLinearCalibrator getAccelerometerCalibrator() {
         return accelerometerCalibrator;
     }
 
     /**
      * Sets accelerometer calibrator to be used to optimize its Mean Square Error (MSE).
      *
      * @param accelerometerCalibrator accelerometer calibrator to be used to optimize its MSE.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if accelerometer calibrator does not use
      *                                  {@link StandardDeviationBodyKinematics} measurements.
      */
     public void setAccelerometerCalibrator(
             final AccelerometerNonLinearCalibrator accelerometerCalibrator) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         if (accelerometerCalibrator != null && accelerometerCalibrator.getMeasurementType()
                 != AccelerometerCalibratorMeasurementType.STANDARD_DEVIATION_BODY_KINEMATICS) {
             throw new IllegalArgumentException();
         }
 
         this.accelerometerCalibrator = accelerometerCalibrator;
     }
 
     /**
      * Gets provided gyroscope calibrator to be used to optimize its Mean Square Error (MSE).
      *
      * @return gyroscope calibrator to be used to optimize its MSE.
      */
     public GyroscopeNonLinearCalibrator getGyroscopeCalibrator() {
         return gyroscopeCalibrator;
     }
 
     /**
      * Sets gyroscope calibrator to be used to optimize its Mean Square Error (MSE).
      *
      * @param gyroscopeCalibrator gyroscope calibrator to be use dto optimize its MSE.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if gyroscope calibrator does not use
      *                                  {@link BodyKinematicsSequence} sequences of
      *                                  {@link StandardDeviationTimedBodyKinematics}.
      */
     public void setGyroscopeCalibrator(
             final GyroscopeNonLinearCalibrator gyroscopeCalibrator) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         if (gyroscopeCalibrator != null && gyroscopeCalibrator.getMeasurementOrSequenceType()
                 != GyroscopeCalibratorMeasurementOrSequenceType.BODY_KINEMATICS_SEQUENCE) {
             throw new IllegalArgumentException();
         }
 
         this.gyroscopeCalibrator = gyroscopeCalibrator;
     }
 
     /**
      * Gets mapper to convert {@link StandardDeviationBodyKinematics} accelerometer measurements
      * into quality scores.
      *
      * @return mapper to convert accelerometer measurements into quality scores.
      */
     public QualityScoreMapper<StandardDeviationBodyKinematics> getAccelerometerQualityScoreMapper() {
         return accelerometerQualityScoreMapper;
     }
 
     /**
      * Sets mapper to convert {@link StandardDeviationBodyKinematics} accelerometer measurements
      * into quality scores.
      *
      * @param accelerometerQualityScoreMapper mapper to convert accelerometer measurements
      *                                        into quality scores.
      * @throws LockedException if optimizer is already running.
      */
     public void setAccelerometerQualityScoreMapper(
             final QualityScoreMapper<StandardDeviationBodyKinematics> accelerometerQualityScoreMapper)
             throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         this.accelerometerQualityScoreMapper = accelerometerQualityScoreMapper;
     }
 
     /**
      * Gets mapper to convert {@link BodyKinematicsSequence} gyroscope sequences of
      * {@link StandardDeviationTimedBodyKinematics} into quality scores.
      *
      * @return mapper to convert gyroscope sequences into quality scores.
      */
     public QualityScoreMapper<BodyKinematicsSequence<StandardDeviationTimedBodyKinematics>>
     getGyroscopeQualityScoreMapper() {
         return gyroscopeQualityScoreMapper;
     }
 
     /**
      * Sets mapper to convert {@link BodyKinematicsSequence} gyroscope sequences of
      * {@link StandardDeviationTimedBodyKinematics} into quality scores.
      *
      * @param gyroscopeQualityScoreMapper mapper to convert gyroscope sequences
      *                                    into quality scores.
      * @throws LockedException if optimizer is already running.
      */
     public void setGyroscopeQualityScoreMapper(
             final QualityScoreMapper<BodyKinematicsSequence<StandardDeviationTimedBodyKinematics>>
                     gyroscopeQualityScoreMapper) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         this.gyroscopeQualityScoreMapper = gyroscopeQualityScoreMapper;
     }
 
     /**
      * Gets rule to convert accelerometer and gyroscope MSE values into a single
      * global MSE value.
      *
      * @return rule to convert accelerometer and gyroscope MSE values into a
      * single global MSE value.
      */
     public AccelerometerAndGyroscopeMseRule getMseRule() {
         return mseRule;
     }
 
     /**
      * Sets rule to convert accelerometer and gyroscope MSE values into a single
      * global MSE value.
      *
      * @param mseRule rule to convert accelerometer and gyroscope MSE values into
      *                a single global MSE value.
      * @throws LockedException if optimizer is already running.
      */
     public void setMseRule(final AccelerometerAndGyroscopeMseRule mseRule) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         this.mseRule = mseRule;
     }
 
     /**
      * Gets the minimum threshold factor.
      *
      * @return minimum threshold factor.
      */
     public double getMinThresholdFactor() {
         return minThresholdFactor;
     }
 
     /**
      * Gets the maximum threshold factor.
      *
      * @return maximum threshold factor.
      */
     public double getMaxThresholdFactor() {
         return maxThresholdFactor;
     }
 
     /**
      * Sets a range of threshold factor values to get an optimized threshold factor value.
      *
      * @param minThresholdFactor minimum threshold.
      * @param maxThresholdFactor maximum threshold.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if either minimum or maximum values are
      *                                  negative, or if the minimum value is larger
      *                                  than the maximum one.
      */
     public void setThresholdFactorRange(final double minThresholdFactor, final double maxThresholdFactor)
             throws LockedException {
         if (running) {
             throw new LockedException();
         }
         if (minThresholdFactor < 0.0 || maxThresholdFactor < 0.0 || minThresholdFactor >= maxThresholdFactor) {
             throw new IllegalArgumentException();
         }
 
         this.minThresholdFactor = minThresholdFactor;
         this.maxThresholdFactor = maxThresholdFactor;
     }
 
     /**
      * Indicates whether this optimizer is ready to start optimization.
      *
      * @return true if this optimizer is ready, false otherwise.
      */
     @Override
     public boolean isReady() {
         return super.isReady() && accelerometerCalibrator != null
                 && gyroscopeCalibrator != null
                 && accelerometerQualityScoreMapper != null
                 && gyroscopeQualityScoreMapper != null
                 && mseRule != null;
     }
 
     /**
      * Gets the time interval between input measurements provided to the
      * {@link #getDataSource()} expressed in seconds (s).
      *
      * @return time interval between input measurements.
      */
     public double getTimeInterval() {
         return generator.getTimeInterval();
     }
 
     /**
      * Sets time interval between input measurements provided to the
      * {@link #getDataSource()} expressed in seconds (s).
      *
      * @param timeInterval time interval between input measurements.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is negative.
      */
     public void setTimeInterval(final double timeInterval) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setTimeInterval(timeInterval);
     }
 
     /**
      * Gets the time interval between input measurements provided to the
      * {@link #getDataSource()}.
      *
      * @return time interval between input measurements.
      */
     public Time getTimeIntervalAsTime() {
         return generator.getTimeIntervalAsTime();
     }
 
     /**
      * Gets the time interval between input measurements provided to the
      * {@link #getDataSource()}.
      *
      * @param result instance where the time interval will be stored.
      */
     public void getTimeIntervalAsTime(final Time result) {
         generator.getTimeIntervalAsTime(result);
     }
 
     /**
      * Sets time interval between input measurements provided to the
      * {@link #getDataSource()}.
      *
      * @param timeInterval time interval between input measurements.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is negative.
      */
     public void setTimeInterval(final Time timeInterval) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setTimeInterval(timeInterval);
     }
 
     /**
      * Gets minimum number of input measurements provided to the
      * {@link #getDataSource()} required in a static interval to be taken
      * into account.
      * Smaller static intervals will be discarded.
      *
      * @return a minimum number of input measurements required in a static interval
      * to be taken into account.
      */
     public int getMinStaticSamples() {
         return generator.getMinStaticSamples();
     }
 
     /**
      * Sets minimum number of input measurements provided to the
      * {@link #getDataSource()} required in a static interval to be taken
      * into account.
      * Smaller static intervals will be discarded.
      *
      * @param minStaticSamples a minimum number of input measurements required in
      *                         a static interval to be taken into account.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is less than 2.
      */
     public void setMinStaticSamples(final int minStaticSamples) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setMinStaticSamples(minStaticSamples);
     }
 
     /**
      * Gets maximum number of input measurements provided to the
      * {@link #getDataSource()} allowed in dynamic intervals.
      *
      * @return maximum number of input measurements allowed in dynamic intervals.
      */
     public int getMaxDynamicSamples() {
         return generator.getMaxDynamicSamples();
     }
 
     /**
      * Sets maximum number of input measurements provided to the
      * {@link #getDataSource()} allowed in dynamic intervals.
      *
      * @param maxDynamicSamples maximum number of input measurements allowed in
      *                          dynamic intervals.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is less than 2.
      */
     public void setMaxDynamicSamples(final int maxDynamicSamples) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setMaxDynamicSamples(maxDynamicSamples);
     }
 
     /**
      * Gets length of number of input measurements provided to the
      * {@link #getDataSource()} to keep within the window being processed
      * to determine instantaneous accelerometer noise level.
      *
      * @return length of input measurements to keep within the window.
      */
     public int getWindowSize() {
         return generator.getWindowSize();
     }
 
     /**
      * Sets length of number of input measurements provided to the
      * {@link #getDataSource()} to keep within the window being processed
      * to determine instantaneous noise level.
      * Window size must always be larger than the allowed minimum value, which is 2
      * and must have an odd value.
      *
      * @param windowSize length of number of samples to keep within the window.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is not valid.
      */
     public void setWindowSize(final int windowSize) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setWindowSize(windowSize);
     }
 
     /**
      * Gets number of input measurements provided to the
      * {@link #getDataSource()} to be processed initially while keeping the sensor
      * static to find the base noise level when the device is static.
      *
      * @return number of samples to be processed initially.
      */
     public int getInitialStaticSamples() {
         return generator.getInitialStaticSamples();
     }
 
     /**
      * Sets number of input parameters provided to the {@link #getDataSource()}
      * to be processed initially while keeping the sensor static to
      * find the base noise level when the device is static.
      *
      * @param initialStaticSamples number of samples ot be processed initially.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is less than
      *                                  {@link TriadStaticIntervalDetector#MINIMUM_INITIAL_STATIC_SAMPLES}.
      */
     public void setInitialStaticSamples(final int initialStaticSamples) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setInitialStaticSamples(initialStaticSamples);
     }
 
     /**
      * Gets factor to determine that a sudden movement has occurred during
      * initialization if instantaneous noise level exceeds accumulated noise
      * level by this factor amount.
      * This factor is unit-less.
      *
      * @return factor to determine that a sudden movement has occurred.
      */
     public double getInstantaneousNoiseLevelFactor() {
         return generator.getInstantaneousNoiseLevelFactor();
     }
 
     /**
      * Sets factor to determine that a sudden movement has occurred during
      * initialization if instantaneous noise level exceeds accumulated noise
      * level by this factor amount.
      * This factor is unit-less.
      *
      * @param instantaneousNoiseLevelFactor factor to determine that a sudden
      *                                      movement has occurred during
      *                                      initialization.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is zero or negative.
      */
     public void setInstantaneousNoiseLevelFactor(final double instantaneousNoiseLevelFactor) throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setInstantaneousNoiseLevelFactor(instantaneousNoiseLevelFactor);
     }
 
     /**
      * Gets the overall absolute threshold to determine whether there has been
      * excessive motion during the whole initialization phase.
      * This threshold is expressed in meters per squared second (m/s^2).
      *
      * @return overall absolute threshold to determine whether there has been
      * excessive motion.
      */
     public double getBaseNoiseLevelAbsoluteThreshold() {
         return generator.getBaseNoiseLevelAbsoluteThreshold();
     }
 
     /**
      * Sets the overall absolute threshold to determine whether there has been
      * excessive motion during the whole initialization phase.
      * This threshold is expressed in meters per squared second (m/s^2).
      *
      * @param baseNoiseLevelAbsoluteThreshold overall absolute threshold to
      *                                        determine whether there has been
      *                                        excessive motion.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is zero or negative.
      */
     public void setBaseNoiseLevelAbsoluteThreshold(final double baseNoiseLevelAbsoluteThreshold)
             throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setBaseNoiseLevelAbsoluteThreshold(baseNoiseLevelAbsoluteThreshold);
     }
 
     /**
      * Gets the overall absolute threshold to determine whether there has been
      * excessive motion during the whole initialization phase.
      *
      * @return overall absolute threshold to determine whether there has been
      * excessive motion.
      */
     public Acceleration getBaseNoiseLevelAbsoluteThresholdAsMeasurement() {
         return generator.getBaseNoiseLevelAbsoluteThresholdAsMeasurement();
     }
 
     /**
      * Gets the overall absolute threshold to determine whether there has been
      * excessive motion during the whole initialization phase.
      *
      * @param result instance where the result will be stored.
      */
     public void getBaseNoiseLevelAbsoluteThresholdAsMeasurement(final Acceleration result) {
         generator.getBaseNoiseLevelAbsoluteThresholdAsMeasurement(result);
     }
 
     /**
      * Sets the overall absolute threshold to determine whether there has been
      * excessive motion during the whole initialization phase.
      *
      * @param baseNoiseLevelAbsoluteThreshold overall absolute threshold to
      *                                        determine whether there has been
      *                                        excessive motion.
      * @throws LockedException          if optimizer is already running.
      * @throws IllegalArgumentException if provided value is zero or negative.
      */
     public void setBaseNoiseLevelAbsoluteThreshold(final Acceleration baseNoiseLevelAbsoluteThreshold)
             throws LockedException {
         if (running) {
             throw new LockedException();
         }
 
         generator.setBaseNoiseLevelAbsoluteThreshold(baseNoiseLevelAbsoluteThreshold);
     }
 
     /**
      * Gets accelerometer base noise level that has been detected during
      * initialization of the best solution that has been found expressed in
      * meters per squared second (m/s^2).
      * This is equal to the standard deviation of the accelerometer measurements
      * during the initialization phase.
      *
      * @return accelerometer base noise level of the best solution that has been
      * found.
      */
     public double getAccelerometerBaseNoiseLevel() {
         return baseNoiseLevel;
     }
 
     /**
      * Gets accelerometer base noise level that has been detected during
      * initialization of the best solution that has been found.
      * This is equal to the standard deviation of the accelerometer measurements
      * during the initialization phase.
      *
      * @return accelerometer base noise level of the best solution that has been
      * found.
      */
     public Acceleration getAccelerometerBaseNoiseLevelAsMeasurement() {
         return createMeasurement(baseNoiseLevel, getDefaultUnit());
     }
 
     /**
      * Gets accelerometer base noise level that has been detected during
      * initialization of the best solution that has been found.
      * This is equal to the standard deviation of the accelerometer measurements
      * during the initialization phase.
      *
      * @param result instance where the result will be stored.
      */
     public void getAccelerometerBaseNoiseLevelAsMeasurement(final Acceleration result) {
         result.setValue(baseNoiseLevel);
         result.setUnit(getDefaultUnit());
     }
 
     /**
      * Gets gyroscope base noise level PSD (Power Spectral Density)
      * expressed in (rad^2/s).
      *
      * @return gyroscope base noise level PSD.
      */
     public double getGyroscopeBaseNoiseLevelPsd() {
         return angularSpeedNoiseRootPsd * angularSpeedNoiseRootPsd;
     }
 
     /**
      * Gets gyroscope base noise level root PSD (Power Spectral Density)
      * expressed in (rad * s^-0.5)
      *
      * @return gyroscope base noise level root PSD.
      */
     @Override
     public double getGyroscopeBaseNoiseLevelRootPsd() {
         return angularSpeedNoiseRootPsd;
     }
 
     /**
      * Gets accelerometer base noise level PSD (Power Spectral Density)
      * expressed in (m^2 * s^-3).
      *
      * @return accelerometer base noise level PSD.
      */
     public double getAccelerometerBaseNoiseLevelPsd() {
         return baseNoiseLevel * baseNoiseLevel * getTimeInterval();
     }
 
     /**
      * Gets accelerometer base noise level root PSD (Power Spectral Density)
      * expressed in (m * s^-1.5).
      *
      * @return accelerometer base noise level root PSD.
      */
     @Override
     public double getAccelerometerBaseNoiseLevelRootPsd() {
         return baseNoiseLevel * Math.sqrt(getTimeInterval());
     }
 
     /**
      * Gets the threshold to determine static/dynamic period changes expressed in
      * meters per squared second (m/s^2) for the best calibration solution that
      * has been found.
      *
      * @return threshold to determine static/dynamic period changes for the best
      * solution.
      */
     public double getThreshold() {
         return threshold;
     }
 
     /**
      * Gets the threshold to determine static/dynamic period changes for the best
      * calibration solution that has been found.
      *
      * @return threshold to determine static/dynamic period changes for the best
      * solution.
      */
     public Acceleration getThresholdAsMeasurement() {
         return createMeasurement(threshold, getDefaultUnit());
     }
 
     /**
      * Get the threshold to determine static/dynamic period changes for the best
      * calibration solution that has been found.
      *
      * @param result instance where the result will be stored.
      */
     public void getThresholdAsMeasurement(final Acceleration result) {
         result.setValue(threshold);
         result.setUnit(getDefaultUnit());
     }
 
     /**
      * Gets estimated standard deviation norm of accelerometer bias expressed in
      * meters per squared second (m/s^2).
      * This can be used as the initial accelerometer bias uncertainty for
      * {@link INSLooselyCoupledKalmanInitializerConfig} or {@link INSTightlyCoupledKalmanInitializerConfig}.
      *
      * @return estimated standard deviation norm of accelerometer bias or null
      * if not available.
      */
     public Double getEstimatedAccelerometerBiasStandardDeviationNorm() {
         return estimatedAccelerometerCovariance != null
                 ? Math.sqrt(getEstimatedAccelerometerBiasFxVariance()
                 + getEstimatedAccelerometerBiasFyVariance()
                 + getEstimatedAccelerometerBiasFzVariance())
                 : null;
     }
 
     /**
      * Gets estimated x coordinate variance of accelerometer bias expressed in (m^2/s^4).
      *
      * @return estimated x coordinate variance of accelerometer bias or null if not available.
      */
     public Double getEstimatedAccelerometerBiasFxVariance() {
         return estimatedAccelerometerCovariance != null
                 ? estimatedAccelerometerCovariance.getElementAt(0, 0) : null;
     }
 
     /**
      * Gets estimated y coordinate variance of accelerometer bias expressed in (m^2/s^4).
      *
      * @return estimated y coordinate variance of accelerometer bias or null if not available.
      */
     public Double getEstimatedAccelerometerBiasFyVariance() {
         return estimatedAccelerometerCovariance != null
                 ? estimatedAccelerometerCovariance.getElementAt(1, 1) : null;
     }
 
     /**
      * Gets estimated z coordinate variance of accelerometer bias expressed in (m^2/s^4).
      *
      * @return estimated z coordinate variance of accelerometer bias or null if not available.
      */
     public Double getEstimatedAccelerometerBiasFzVariance() {
         return estimatedAccelerometerCovariance != null
                 ? estimatedAccelerometerCovariance.getElementAt(2, 2) : null;
     }
 
     /**
      * Gets an array containing x,y,z components of estimated accelerometer biases
      * expressed in meters per squared second (m/s^2).
      *
      * @return array containing x,y,z components of estimated accelerometer biases.
      */
     public double[] getEstimatedAccelerometerBiases() {
         return estimatedAccelerometerBiases;
     }
 
     /**
      * Gets estimated accelerometer scale factors and cross-coupling errors.
      * This is the product of matrix Ta containing cross-coupling errors and Ka
      * containing scaling factors.
      * So that:
      * <pre>
      *     Ma = [sx    mxy  mxz] = Ta*Ka
      *          [myx   sy   myz]
      *          [mzx   mzy  sz ]
      * </pre>
      * Where:
      * <pre>
      *     Ka = [sx 0   0 ]
      *          [0  sy  0 ]
      *          [0  0   sz]
      * </pre>
      * and
      * <pre>
      *     Ta = [1          -alphaXy    alphaXz ]
      *          [alphaYx    1           -alphaYz]
      *          [-alphaZx   alphaZy     1       ]
      * </pre>
      * Hence:
      * <pre>
      *     Ma = [sx    mxy  mxz] = Ta*Ka =  [sx             -sy * alphaXy   sz * alphaXz ]
      *          [myx   sy   myz]            [sx * alphaYx   sy              -sz * alphaYz]
      *          [mzx   mzy  sz ]            [-sx * alphaZx  sy * alphaZy    sz           ]
      * </pre>
      * This instance allows any 3x3 matrix. However, typically alphaYx, alphaZx and alphaZy
      * are considered to be zero if the accelerometer z-axis is assumed to be the same
      * as the body z-axis. When this is assumed, myx = mzx = mzy = 0 and the Ma matrix
      * becomes upper diagonal:
      * <pre>
      *     Ma = [sx    mxy  mxz]
      *          [0     sy   myz]
      *          [0     0    sz ]
      * </pre>
      * Values of this matrix are unit-less.
      *
      * @return estimated accelerometer scale factors and cross-coupling errors, or null
      * if not available.
      */
     public Matrix getEstimatedAccelerometerMa() {
         return estimatedAccelerometerMa;
     }
 
     /**
      * Gets estimated standard deviation norm of gyroscope bias expressed in
      * radians per second (rad/s).
      * This can be used as the initial gyroscope bias uncertainty for
      * {@link INSLooselyCoupledKalmanInitializerConfig} or {@link INSTightlyCoupledKalmanInitializerConfig}.
      *
      * @return estimated standard deviation norm of gyroscope bias or null
      * if not available.
      */
     public Double getEstimatedGyroscopeBiasStandardDeviationNorm() {
         return estimatedGyroscopeCovariance != null
                 ? Math.sqrt(getEstimatedGyroscopeBiasXVariance()
                 + getEstimatedGyroscopeBiasYVariance()
                 + getEstimatedGyroscopeBiasZVariance())
                 : null;
     }
 
     /**
      * Gets estimated x coordinate variance of gyroscope bias expressed in (rad^2/s^2).
      *
      * @return estimated x coordinate variance of gyroscope bias or null if not available.
      */
     public Double getEstimatedGyroscopeBiasXVariance() {
         return estimatedGyroscopeCovariance != null
                 ? estimatedGyroscopeCovariance.getElementAt(0, 0) : null;
     }
 
     /**
      * Gets estimated y coordinate variance of gyroscope bias expressed in (rad^2/s^2).
      *
      * @return estimated y coordinate variance of gyroscope bias or null if not available.
      */
     public Double getEstimatedGyroscopeBiasYVariance() {
         return estimatedGyroscopeCovariance != null
                 ? estimatedGyroscopeCovariance.getElementAt(1, 1) : null;
     }
 
     /**
      * Gets estimated z coordinate variance of gyroscope bias expressed in (rad^2/s^2).
      *
      * @return estimated z coordinate variance of gyroscope bias or null if not available.
      */
     public Double getEstimatedGyroscopeBiasZVariance() {
         return estimatedGyroscopeCovariance != null
                 ? estimatedGyroscopeCovariance.getElementAt(2, 2) : null;
     }
 
     /**
      * Gets the array containing x,y,z components of estimated gyroscope biases
      * expressed in radians per second (rad/s).
      *
      * @return array containing x,y,z components of estimated gyroscope biases.
      */
     public double[] getEstimatedGyroscopeBiases() {
         return estimatedGyroscopeBiases;
     }
 
     /**
      * Gets estimated gyroscope scale factors and cross-coupling errors.
      * This is the product of matrix Tg containing cross-coupling errors and Kg
      * containing scaling factors.
      * So that:
      * <pre>
      *     Mg = [sx    mxy  mxz] = Tg*Kg
      *          [myx   sy   myz]
      *          [mzx   mzy  sz ]
      * </pre>
      * Where:
      * <pre>
      *     Kg = [sx 0   0 ]
      *          [0  sy  0 ]
      *          [0  0   sz]
      * </pre>
      * and
      * <pre>
      *     Tg = [1          -alphaXy    alphaXz ]
      *          [alphaYx    1           -alphaYz]
      *          [-alphaZx   alphaZy     1       ]
      * </pre>
      * Hence:
      * <pre>
      *     Mg = [sx    mxy  mxz] = Tg*Kg =  [sx             -sy * alphaXy   sz * alphaXz ]
      *          [myx   sy   myz]            [sx * alphaYx   sy              -sz * alphaYz]
      *          [mzx   mzy  sz ]            [-sx * alphaZx  sy * alphaZy    sz           ]
      * </pre>
      * This instance allows any 3x3 matrix. However, typically alphaYx, alphaZx and alphaZy
      * are considered to be zero if the gyroscope z-axis is assumed to be the same
      * as the body z-axis. When this is assumed, myx = mzx = mzy = 0 and the Mg matrix
      * becomes upper diagonal:
      * <pre>
      *     Mg = [sx    mxy  mxz]
      *          [0     sy   myz]
      *          [0     0    sz ]
      * </pre>
      * Values of this matrix are unit-less.
      *
      * @return estimated gyroscope scale factors and cross-coupling errors.
      */
     public Matrix getEstimatedGyroscopeMg() {
         return estimatedGyroscopeMg;
     }
 
     /**
      * Gets estimated G-dependent cross-biases introduced on the gyroscope by the
      * specific forces sensed by the accelerometer.
      *
      * @return a 3x3 matrix containing g-dependent cross biases.
      */
     public Matrix getEstimatedGyroscopeGg() {
         return estimatedGyroscopeGg;
     }
 
     /**
      * Evaluates calibration Mean Square Error (MSE) for the provided threshold factor.
      *
      * @param thresholdFactor threshold factor to be used for interval detection
      *                        and measurement generation to be used for
      *                        calibration.
      * @return calibration MSE.
      * @throws LockedException                                   if the generator is busy.
      * @throws CalibrationException                              if calibration fails.
      * @throws NotReadyException                                 if the calibrator is not ready.
      * @throws IntervalDetectorThresholdFactorOptimizerException interval detection failed.
      */
     protected double evaluateForThresholdFactor(final double thresholdFactor) throws LockedException,
             CalibrationException, NotReadyException, IntervalDetectorThresholdFactorOptimizerException {
         if (measurements == null) {
             measurements = new ArrayList<>();
         } else {
             measurements.clear();
         }
 
         if (sequences == null) {
             sequences = new ArrayList<>();
         } else {
             sequences.clear();
         }
 
         generator.reset();
         generator.setThresholdFactor(thresholdFactor);
 
         var count = dataSource.count();
         var failed = false;
         for (var i = 0; i < count; i++) {
             final var timedBodyKinematics = dataSource.getAt(i);
             if (!generator.process(timedBodyKinematics)) {
                 failed = true;
                 break;
             }
         }
 
         if (failed || generator.getStatus() == TriadStaticIntervalDetector.Status.FAILED) {
             // interval detection failed
             return Double.MAX_VALUE;
         }
 
         // check that enough measurements have been obtained
         if (measurements.size() < accelerometerCalibrator.getMinimumRequiredMeasurements()
                 || sequences.size() < gyroscopeCalibrator.getMinimumRequiredMeasurementsOrSequences()) {
             return Double.MAX_VALUE;
         }
 
         // set calibrator measurements
         switch (accelerometerCalibrator.getMeasurementType()) {
             case STANDARD_DEVIATION_BODY_KINEMATICS:
                 if (accelerometerCalibrator.isOrderedMeasurementsRequired()) {
                     final var calibrator =
                             (OrderedStandardDeviationBodyKinematicsAccelerometerCalibrator) accelerometerCalibrator;
 
                     calibrator.setMeasurements(measurements);
 
                 } else {
                     final var calibrator =
                             (UnorderedStandardDeviationBodyKinematicsAccelerometerCalibrator) accelerometerCalibrator;
 
                     calibrator.setMeasurements(measurements);
                 }
 
                 if (accelerometerCalibrator.isQualityScoresRequired()) {
                     final var calibrator = (QualityScoredAccelerometerCalibrator) accelerometerCalibrator;
 
                     final var size = measurements.size();
                     final var qualityScores = new double[size];
                     for (var i = 0; i < size; i++) {
                         qualityScores[i] = accelerometerQualityScoreMapper.map(measurements.get(i));
                     }
                     calibrator.setQualityScores(qualityScores);
                 }
                 break;
             case FRAME_BODY_KINEMATICS, STANDARD_DEVIATION_FRAME_BODY_KINEMATICS:
                 // Throw exception. Cannot use frames
             default:
                 throw new IntervalDetectorThresholdFactorOptimizerException();
         }
 
         if (gyroscopeCalibrator.getMeasurementOrSequenceType()
                 == GyroscopeCalibratorMeasurementOrSequenceType.BODY_KINEMATICS_SEQUENCE) {
             final var calibrator = (OrderedBodyKinematicsSequenceGyroscopeCalibrator) gyroscopeCalibrator;
 
             calibrator.setSequences(sequences);
         } else {
             throw new IntervalDetectorThresholdFactorOptimizerException();
         }
 
         if (gyroscopeCalibrator.isQualityScoresRequired()) {
             final var calibrator = (QualityScoredGyroscopeCalibrator) gyroscopeCalibrator;
 
             final var size = sequences.size();
             final var qualityScores = new double[size];
             for (var i = 0; i < size; i++) {
                 qualityScores[i] = gyroscopeQualityScoreMapper.map(sequences.get(i));
             }
             calibrator.setQualityScores(qualityScores);
         }
 
         accelerometerCalibrator.calibrate();
 
         // once we have accelerometer estimations, we can set known
         // accelerometer bias and cross-coupling errors to the gyroscope calibrator
         if (gyroscopeCalibrator instanceof AccelerometerDependentGyroscopeCalibrator accelGyroCalibrator) {
 
             final double[] bias;
             if (accelerometerCalibrator instanceof UnknownBiasAccelerometerCalibrator unknownBiasAccelerometerCalibrator) {
                 bias = unknownBiasAccelerometerCalibrator.getEstimatedBiases();
 
             } else if (accelerometerCalibrator instanceof KnownBiasAccelerometerCalibrator knownBiasAccelerometerCalibrator) {
                 bias = knownBiasAccelerometerCalibrator.getBias();
             } else {
                 bias = null;
             }
 
             if (bias != null) {
                 accelGyroCalibrator.setAccelerometerBias(bias);
                 accelGyroCalibrator.setAccelerometerMa(accelerometerCalibrator.getEstimatedMa());
             }
         }
 
         gyroscopeCalibrator.calibrate();
 
         final var accelerometerMse = accelerometerCalibrator.getEstimatedMse();
         final var gyroscopeMse = gyroscopeCalibrator.getEstimatedMse();
 
         // convert accelerometer and gyroscope mse to global mse
         final var mse = mseRule.evaluate(accelerometerMse, gyroscopeMse);
         if (mse < minMse) {
             keepBestResult(mse, thresholdFactor);
         }
         return mse;
     }
 
     /**
      * Initializes accelerometer + gyroscope measurement generator to convert timed
      * body kinematics measurements after interval detection into measurements and
      * sequences used for accelerometer and gyroscope calibration.
      */
     private void initialize() {
         final var generatorListener = new AccelerometerAndGyroscopeMeasurementsGeneratorListener() {
             @Override
             public void onInitializationStarted(final AccelerometerAndGyroscopeMeasurementsGenerator generator) {
                 // not needed
             }
 
             @Override
             public void onInitializationCompleted(
                     final AccelerometerAndGyroscopeMeasurementsGenerator generator,
                     final double accelerometerBaseNoiseLevel) {
                 // not needed
             }
 
             @Override
             public void onError(
                     final AccelerometerAndGyroscopeMeasurementsGenerator generator,
                     final TriadStaticIntervalDetector.ErrorReason reason) {
                 // not needed
             }
 
             @Override
             public void onStaticIntervalDetected(final AccelerometerAndGyroscopeMeasurementsGenerator generator) {
                 // not needed
             }
 
             @Override
             public void onDynamicIntervalDetected(final AccelerometerAndGyroscopeMeasurementsGenerator generator) {
                 // not needed
             }
 
             @Override
             public void onStaticIntervalSkipped(final AccelerometerAndGyroscopeMeasurementsGenerator generator) {
                 // not needed
             }
 
             @Override
             public void onDynamicIntervalSkipped(final AccelerometerAndGyroscopeMeasurementsGenerator generator) {
                 // not needed
             }
 
             @Override
             public void onGeneratedAccelerometerMeasurement(
                     final AccelerometerAndGyroscopeMeasurementsGenerator generator,
                     final StandardDeviationBodyKinematics measurement) {
                 measurements.add(measurement);
             }
 
             @Override
             public void onGeneratedGyroscopeMeasurement(
                     final AccelerometerAndGyroscopeMeasurementsGenerator generator,
                     final BodyKinematicsSequence<StandardDeviationTimedBodyKinematics> sequence) {
                 sequences.add(sequence);
             }
 
             @Override
             public void onReset(final AccelerometerAndGyroscopeMeasurementsGenerator generator) {
                 // not needed
             }
         };
 
         generator = new AccelerometerAndGyroscopeMeasurementsGenerator(generatorListener);
     }
 
     /**
      * Keeps the best calibration solution found so far.
      *
      * @param mse             Estimated Mean Square Error during calibration.
      * @param thresholdFactor threshold factor to be kept.
      */
     private void keepBestResult(final double mse, final double thresholdFactor) {
         minMse = mse;
         optimalThresholdFactor = thresholdFactor;
 
         angularSpeedNoiseRootPsd = generator.getGyroscopeBaseNoiseLevelRootPsd();
         baseNoiseLevel = generator.getAccelerometerBaseNoiseLevel();
         threshold = generator.getThreshold();
 
         if (estimatedAccelerometerCovariance == null) {
             estimatedAccelerometerCovariance = new Matrix(accelerometerCalibrator.getEstimatedCovariance());
         } else {
             estimatedAccelerometerCovariance.copyFrom(accelerometerCalibrator.getEstimatedCovariance());
         }
         if (estimatedAccelerometerMa == null) {
             estimatedAccelerometerMa = new Matrix(accelerometerCalibrator.getEstimatedMa());
         } else {
             estimatedAccelerometerMa.copyFrom(accelerometerCalibrator.getEstimatedMa());
         }
         if (accelerometerCalibrator instanceof UnknownBiasAccelerometerCalibrator unknownBiasAccelerometerCalibrator) {
             estimatedAccelerometerBiases = unknownBiasAccelerometerCalibrator.getEstimatedBiases();
         } else if (accelerometerCalibrator instanceof
                 KnownBiasAccelerometerCalibrator knownBiasAccelerometerCalibrator) {
             estimatedAccelerometerBiases = knownBiasAccelerometerCalibrator.getBias();
         }
 
         if (estimatedGyroscopeCovariance == null) {
             estimatedGyroscopeCovariance = new Matrix(gyroscopeCalibrator.getEstimatedCovariance());
         } else {
             estimatedGyroscopeCovariance.copyFrom(gyroscopeCalibrator.getEstimatedCovariance());
         }
         if (estimatedGyroscopeMg == null) {
             estimatedGyroscopeMg = new Matrix(gyroscopeCalibrator.getEstimatedMg());
         } else {
             estimatedGyroscopeMg.copyFrom(gyroscopeCalibrator.getEstimatedMg());
         }
         if (estimatedGyroscopeGg == null) {
             estimatedGyroscopeGg = new Matrix(gyroscopeCalibrator.getEstimatedGg());
         } else {
             estimatedGyroscopeGg.copyFrom(gyroscopeCalibrator.getEstimatedGg());
         }
         if (gyroscopeCalibrator instanceof UnknownBiasGyroscopeCalibrator unknownBiasGyroscopeCalibrator) {
             estimatedGyroscopeBiases = unknownBiasGyroscopeCalibrator.getEstimatedBiases();
         } else if (gyroscopeCalibrator instanceof KnownBiasAccelerometerCalibrator knownBiasAccelerometerCalibrator) {
             estimatedGyroscopeBiases = knownBiasAccelerometerCalibrator.getBias();
         }
     }
 
     /**
      * Creates an acceleration instance using the provided value and unit.
      *
      * @param value value of measurement.
      * @param unit  unit of value.
      * @return created acceleration.
      */
     private Acceleration createMeasurement(final double value, final AccelerationUnit unit) {
         return new Acceleration(value, unit);
     }
 
     /**
      * Gets the default unit for acceleration, which is meters per
      * squared second (m/s^2).
      *
      * @return default unit for acceleration.
      */
     private AccelerationUnit getDefaultUnit() {
         return AccelerationUnit.METERS_PER_SQUARED_SECOND;
     }
 }