MRPT logo

mrpt::slam::CBeaconMap Class Reference

A class for storing a map of 3D probabilistic beacons, using a Montecarlo, Gaussian, or Sum of Gaussians (SOG) representation (for range-only SLAM). More...

#include <mrpt/slam/CBeaconMap.h>

Inheritance diagram for mrpt::slam::CBeaconMap:

mrpt::slam::CMetricMap mrpt::utils::CSerializable

List of all members.

Classes

struct  TInsertionOptions
 This struct contains data for choosing the method by which new beacons are inserted in the map. More...
struct  TLikelihoodOptions
 With this struct options are provided to the likelihood computations. More...

Public Member Functions

 CBeaconMap ()
 Constructor.
virtual ~CBeaconMap ()
 Virtual destructor.
float compute3DMatchingRatio (const CMetricMap *otherMap, const CPose3D &otherMapPose, float minDistForCorr=0.10f, float minMahaDistForCorr=2.0f) const
 Computes the ratio in [0,1] of correspondences between "this" and the "otherMap" map, whose 6D pose relative to "this" is "otherMapPose" In the case of a multi-metric map, this returns the average between the maps.
bool saveToMATLABScript3D (std::string file, const char *style="b", float confInterval=0.95f) const
 Save to a MATLAB script which displays 3D error ellipses for the map.
void clear ()
 Clear the map, erasing all landmarks.
size_t size () const
 Returns the stored landmarks count.
bool insertObservation (const CObservation *obs, const CPose3D *robotPose=NULL)
 Insert the observation information into this map.
double computeObservationLikelihood (const CObservation *obs, const CPose3D &takenFrom)
 Computes the (logarithmic) likelihood that a given observation was taken from a given pose in the world being modeled with this map.
void computeMatchingWith2D (const CMetricMap *otherMap, const CPose2D &otherMapPose, float maxDistForCorrespondence, float maxAngularDistForCorrespondence, const CPose2D &angularDistPivotPoint, TMatchingPairList &correspondences, float &correspondencesRatio, float *sumSqrDist=NULL, bool onlyKeepTheClosest=false, bool onlyUniqueRobust=false) const
 Computes the matchings between this and another 2D points map.
void computeMatchingWith3DLandmarks (const mrpt::slam::CBeaconMap *otherMap, TMatchingPairList &correspondences, float &correspondencesRatio, std::vector< bool > &otherCorrespondences) const
 Perform a search for correspondences between "this" and another lansmarks map: Firsly, the landmarks' descriptor is used to find correspondences, then inconsistent ones removed by looking at their 3D poses.
void changeCoordinatesReference (const CPose3D &newOrg)
 Changes the reference system of the map to a given 3D pose.
void changeCoordinatesReference (const CPose3D &newOrg, const mrpt::slam::CBeaconMap *otherMap)
 Changes the reference system of the map "otherMap" and save the result in "this" map.
bool isEmpty () const
 Returns true if the map is empty/no observation has been inserted.
void simulateBeaconReadings (const CPose3D &in_robotPose, const CPoint3D &in_sensorLocationOnRobot, CObservationBeaconRanges &out_Observations) const
 Simulates a reading toward each of the beacons in the landmarks map, if any.
void saveMetricMapRepresentationToFile (const std::string &filNamePrefix) const
 This virtual method saves the map to a file "filNamePrefix"+< some_file_extension >, as an image or in any other applicable way (Notice that other methods to save the map may be implemented in classes implementing this virtual interface).
void saveToTextFile (const std::string &fil) const
 Save a text file with a row per beacon, containing this 11 elements:
  • X Y Z: Mean values
  • VX VY VZ: Variances of each dimension (C11, C22, C33)
  • DET2D DET3D: Determinant of the 2D and 3D covariance matrixes.

void getAs3DObject (mrpt::opengl::CSetOfObjectsPtr &outObj) const
 Returns a 3D object representing the map.
void auxParticleFilterCleanUp ()
 This method is called at the end of each "prediction-update-map insertion" cycle within "mrpt::slam::CMetricMapBuilderRBPF::processActionObservation".
const CBeacongetBeaconByID (CBeacon::TBeaconID id) const
 Returns a pointer to the beacon with the given ID, or NULL if it does not exist.
CBeacongetBeaconByID (CBeacon::TBeaconID id)
 Returns a pointer to the beacon with the given ID, or NULL if it does not exist.

Public Attributes

TSequenceBeacons m_beacons
 The individual beacons:.
mrpt::slam::CBeaconMap::TLikelihoodOptions likelihoodOptions
 With this struct options are provided to the likelihood computations.
mrpt::slam::CBeaconMap::TInsertionOptions insertionOptions
 This struct contains data for choosing the method by which new beacons are inserted in the map.


Detailed Description

A class for storing a map of 3D probabilistic beacons, using a Montecarlo, Gaussian, or Sum of Gaussians (SOG) representation (for range-only SLAM).


The individual beacons are defined as mrpt::slam::CBeacon objects.
When invoking CBeaconMap::insertObservation(), landmarks will be extracted and fused into the map. The only currently supported observation type is mrpt::slam::CObservationBeaconRanges. See insertionOptions and likelihoodOptions for parameters used when creating and fusing beacon landmarks.
Use "TInsertionOptions::insertAsMonteCarlo" to select between 2 different behaviors:

Refer to the papers: []

See also:
CMetricMap

Definition at line 69 of file CBeaconMap.h.


Constructor & Destructor Documentation

mrpt::slam::CBeaconMap::CBeaconMap (  ) 

Constructor.

virtual mrpt::slam::CBeaconMap::~CBeaconMap (  )  [virtual]

Virtual destructor.


Member Function Documentation

void mrpt::slam::CBeaconMap::auxParticleFilterCleanUp (  )  [virtual]

This method is called at the end of each "prediction-update-map insertion" cycle within "mrpt::slam::CMetricMapBuilderRBPF::processActionObservation".

This method should normally do nothing, but in some cases can be used to free auxiliary cached variables.

Implements mrpt::slam::CMetricMap.

void mrpt::slam::CBeaconMap::changeCoordinatesReference ( const CPose3D newOrg,
const mrpt::slam::CBeaconMap otherMap 
)

Changes the reference system of the map "otherMap" and save the result in "this" map.

void mrpt::slam::CBeaconMap::changeCoordinatesReference ( const CPose3D newOrg  ) 

Changes the reference system of the map to a given 3D pose.

void mrpt::slam::CBeaconMap::clear (  )  [virtual]

Clear the map, erasing all landmarks.

Implements mrpt::slam::CMetricMap.

float mrpt::slam::CBeaconMap::compute3DMatchingRatio ( const CMetricMap otherMap,
const CPose3D otherMapPose,
float  minDistForCorr = 0.10f,
float  minMahaDistForCorr = 2.0f 
) const [virtual]

Computes the ratio in [0,1] of correspondences between "this" and the "otherMap" map, whose 6D pose relative to "this" is "otherMapPose" In the case of a multi-metric map, this returns the average between the maps.

This method always return 0 for grid maps.

Parameters:
otherMap [IN] The other map to compute the matching with.
otherMapPose [IN] The 6D pose of the other map as seen from "this".
minDistForCorr [IN] The minimum distance between 2 non-probabilistic map elements for counting them as a correspondence.
minMahaDistForCorr [IN] The minimum Mahalanobis distance between 2 probabilistic map elements for counting them as a correspondence.
Returns:
The matching ratio [0,1]
See also:
computeMatchingWith2D

Implements mrpt::slam::CMetricMap.

void mrpt::slam::CBeaconMap::computeMatchingWith2D ( const CMetricMap otherMap,
const CPose2D otherMapPose,
float  maxDistForCorrespondence,
float  maxAngularDistForCorrespondence,
const CPose2D angularDistPivotPoint,
TMatchingPairList correspondences,
float &  correspondencesRatio,
float *  sumSqrDist = NULL,
bool  onlyKeepTheClosest = false,
bool  onlyUniqueRobust = false 
) const [virtual]

Computes the matchings between this and another 2D points map.

This includes finding:

  • The set of points pairs in each map
  • The mean squared distance between corresponding pairs. This method is the most time critical one into the ICP algorithm.

Parameters:
otherMap [IN] The other map to compute the matching with.
otherMapPose [IN] The pose of the other map as seen from "this".
maxDistForCorrespondence [IN] Maximum 2D linear distance between two points to be matched.
maxAngularDistForCorrespondence [IN] In radians: The aim is to allow larger distances to more distant correspondences.
angularDistPivotPoint [IN] The point used to calculate distances from in both maps.
correspondences [OUT] The detected matchings pairs.
correspondencesRatio [OUT] The ratio [0,1] of points in otherMap with at least one correspondence.
sumSqrDist [OUT] The sum of all matched points squared distances.If undesired, set to NULL, as default.
covariance [OUT] The resulting matching covariance 3x3 matrix, or NULL if undesired.
onlyKeepTheClosest [IN] If set to true, only the closest correspondence will be returned. If false (default) all are returned.
See also:
compute3DMatchingRatio

Reimplemented from mrpt::slam::CMetricMap.

void mrpt::slam::CBeaconMap::computeMatchingWith3DLandmarks ( const mrpt::slam::CBeaconMap otherMap,
TMatchingPairList correspondences,
float &  correspondencesRatio,
std::vector< bool > &  otherCorrespondences 
) const

Perform a search for correspondences between "this" and another lansmarks map: Firsly, the landmarks' descriptor is used to find correspondences, then inconsistent ones removed by looking at their 3D poses.

Parameters:
otherMap [IN] The other map.
correspondences [OUT] The matched pairs between maps.
correspondencesRatio [OUT] This is NumberOfMatchings / NumberOfLandmarksInTheAnotherMap
otherCorrespondences [OUT] Will be returned with a vector containing "true" for the indexes of the other map's landmarks with a correspondence.

double mrpt::slam::CBeaconMap::computeObservationLikelihood ( const CObservation obs,
const CPose3D takenFrom 
) [virtual]

Computes the (logarithmic) likelihood that a given observation was taken from a given pose in the world being modeled with this map.

In the current implementation, this method behaves in a different way according to the nature of the observation's class:

  • "mrpt::slam::CObservation2DRangeScan": This calls "computeLikelihood_RSLC_2007".
  • "mrpt::slam::CObservationStereoImages": This calls "computeLikelihood_SIFT_LandmarkMap".

Parameters:
takenFrom The robot's pose the observation is supposed to be taken from.
obs The observation.
Returns:
This method returns a likelihood value > 0.
See also:
Used in particle filter algorithms, see: CMultiMetricMapPDF::update

Implements mrpt::slam::CMetricMap.

void mrpt::slam::CBeaconMap::getAs3DObject ( mrpt::opengl::CSetOfObjectsPtr &  outObj  )  const [virtual]

Returns a 3D object representing the map.

Implements mrpt::slam::CMetricMap.

CBeacon* mrpt::slam::CBeaconMap::getBeaconByID ( CBeacon::TBeaconID  id  ) 

Returns a pointer to the beacon with the given ID, or NULL if it does not exist.

const CBeacon* mrpt::slam::CBeaconMap::getBeaconByID ( CBeacon::TBeaconID  id  )  const

Returns a pointer to the beacon with the given ID, or NULL if it does not exist.

bool mrpt::slam::CBeaconMap::insertObservation ( const CObservation obs,
const CPose3D robotPose = NULL 
) [virtual]

Insert the observation information into this map.

This method must be implemented in derived classes.

Parameters:
obs The observation
robotPose The 3D pose of the robot mobile base in the map reference system, or NULL (default) if you want to use CPose2D(0,0,deg)
See also:
CObservation::insertObservationInto

Implements mrpt::slam::CMetricMap.

bool mrpt::slam::CBeaconMap::isEmpty (  )  const [virtual]

Returns true if the map is empty/no observation has been inserted.

Implements mrpt::slam::CMetricMap.

void mrpt::slam::CBeaconMap::saveMetricMapRepresentationToFile ( const std::string &  filNamePrefix  )  const [virtual]

This virtual method saves the map to a file "filNamePrefix"+< some_file_extension >, as an image or in any other applicable way (Notice that other methods to save the map may be implemented in classes implementing this virtual interface).

In the case of this class, these files are generated:

  • "filNamePrefix"+"_3D.m": A script for MATLAB for drawing landmarks as 3D ellipses.
  • "filNamePrefix"+"_3D.3DScene": A 3D scene with a "ground plane grid" and the set of ellipsoids in 3D.
  • "filNamePrefix"+"_covs.m": A textual representation (see saveToTextFile)

Implements mrpt::slam::CMetricMap.

bool mrpt::slam::CBeaconMap::saveToMATLABScript3D ( std::string  file,
const char *  style = "b",
float  confInterval = 0.95f 
) const

Save to a MATLAB script which displays 3D error ellipses for the map.

Parameters:
file The name of the file to save the script to.
style The MATLAB-like string for the style of the lines (see 'help plot' in MATLAB for posibilities)
stdCount The ellipsoids will be drawn from the center to a given confidence interval in [0,1], e.g. 2 sigmas=0.95 (default is 2std = 0.95 confidence intervals)
Returns:
Returns false if any error occured, true elsewere.

void mrpt::slam::CBeaconMap::saveToTextFile ( const std::string &  fil  )  const

Save a text file with a row per beacon, containing this 11 elements:

  • X Y Z: Mean values
  • VX VY VZ: Variances of each dimension (C11, C22, C33)
  • DET2D DET3D: Determinant of the 2D and 3D covariance matrixes.

  • C12, C13, C23: Cross covariances

void mrpt::slam::CBeaconMap::simulateBeaconReadings ( const CPose3D in_robotPose,
const CPoint3D in_sensorLocationOnRobot,
CObservationBeaconRanges out_Observations 
) const

Simulates a reading toward each of the beacons in the landmarks map, if any.

Parameters:
in_robotPose This robot pose is used to simulate the ranges to each beacon.
in_sensorLocationOnRobot The 3D position of the sensor on the robot
out_Observations The results will be stored here. NOTICE that the fields "CObservationBeaconRanges::minSensorDistance","CObservationBeaconRanges::maxSensorDistance" and "CObservationBeaconRanges::stdError" MUST BE FILLED OUT before calling this function. An observation will be generated for each beacon in the map, but notice that some of them may be missed if out of the sensor maximum range.

size_t mrpt::slam::CBeaconMap::size (  )  const

Returns the stored landmarks count.


Member Data Documentation

This struct contains data for choosing the method by which new beacons are inserted in the map.

With this struct options are provided to the likelihood computations.

The individual beacons:.

Definition at line 77 of file CBeaconMap.h.




Page generated by Doxygen 1.5.8 for MRPT 0.6.5 SVN: at Thu Feb 26 02:07:47 EST 2009