Detailed Description

An implementation of rejection sampling for generating 2D robot pose from range-only measurements within a landmarks (beacons) map.

Before calling the method "rejectionSampling" to generate the samples, you must call "setParams". It is assumed a planar scenario, where the robot is at a fixed height (default=0).

See also:: bayes::CRejectionSamplingCapable

#include <mrpt/slam/CRejectionSamplingRangeOnlyLocalization.h>

Inheritance diagram for mrpt::slam::CRejectionSamplingRangeOnlyLocalization:

[legend]

List of all members.

Classes
struct	TDataPerBeacon
	Data for each beacon observation with a correspondence with the map. More...
Public Types
typedef CProbabilityParticle < TStateSpace >	TParticle
Public Member Functions
	CRejectionSamplingRangeOnlyLocalization ()
	Constructor.
virtual	~CRejectionSamplingRangeOnlyLocalization ()
	Destructor.
bool	setParams (const CLandmarksMap &beaconsMap, const CObservationBeaconRanges &observation, float sigmaRanges, const CPose2D &oldPose, float robot_z=0, bool autoCheckAngleRanges=true)
	The parameters used in the generation of random samples:
void	rejectionSampling (size_t desiredSamples, std::vector< TParticle > &outSamples, size_t timeoutTrials=1000)
	Generates a set of N independent samples via rejection sampling.
Protected Member Functions
void	RS_drawFromProposal (CPose2D &outSample)
	Generates one sample, drawing from some proposal distribution.
double	RS_observationLikelihood (const CPose2D &x)
	Returns the NORMALIZED observation likelihood (linear, not exponential!!!) at a given point of the state space (values in the range [0,1]).
virtual void	RS_drawFromProposal (TStateSpace &outSample)=0
	Generates one sample, drawing from some proposal distribution.
virtual double	RS_observationLikelihood (const TStateSpace &x)=0
	Returns the NORMALIZED observation likelihood (linear, not exponential!!!) at a given point of the state space (values in the range [0,1]).
Protected Attributes
float	m_z_robot
	Z coordinate of the robot.
float	m_sigmaRanges
CPose2D	m_oldPose
size_t	m_drawIndex
	The index in "m_dataPerBeacon" used to draw samples (the rest will be used to evaluate the likelihood)
std::deque< TDataPerBeacon >	m_dataPerBeacon
	Data for each beacon observation with a correspondence with the map.

Member Typedef Documentation

typedef CProbabilityParticle<TStateSpace> mrpt::bayes::CRejectionSamplingCapable::TParticle [inherited]

Definition at line 50 of file CRejectionSamplingCapable.h.

Constructor & Destructor Documentation

mrpt::slam::CRejectionSamplingRangeOnlyLocalization::CRejectionSamplingRangeOnlyLocalization ( )

Constructor.

virtual mrpt::slam::CRejectionSamplingRangeOnlyLocalization::~CRejectionSamplingRangeOnlyLocalization ( ) [inline, virtual]

Destructor.

Definition at line 64 of file CRejectionSamplingRangeOnlyLocalization.h.

Member Function Documentation

void mrpt::bayes::CRejectionSamplingCapable::rejectionSampling	(	size_t	desiredSamples,
		std::vector< TParticle > &	outSamples,
		size_t	timeoutTrials = `1000`
	)		`[inline, inherited]`

Generates a set of N independent samples via rejection sampling.

Parameters:

desiredSamples	The number of desired samples to generate
outSamples	The output samples.
timeoutTrials	The maximum number of rejection trials for each generated sample (i.e. the maximum number of iterations). This can be used to set a limit to the time complexity of the algorithm for difficult probability densities. All will have equal importance weights (a property of rejection sampling), although those samples generated at timeout will have a different importance weights.

Definition at line 65 of file CRejectionSamplingCapable.h.

References MRPT_START, mrpt::bayes::CRejectionSamplingCapable::RS_drawFromProposal(), mrpt::bayes::CRejectionSamplingCapable::RS_observationLikelihood(), ASSERT_, mrpt::random::randomGenerator, and MRPT_END.

void mrpt::slam::CRejectionSamplingRangeOnlyLocalization::RS_drawFromProposal ( CPose2D & outSample ) [protected]

Generates one sample, drawing from some proposal distribution.

virtual void mrpt::bayes::CRejectionSamplingCapable::RS_drawFromProposal ( TStateSpace & outSample ) [protected, pure virtual, inherited]

Generates one sample, drawing from some proposal distribution.

double mrpt::slam::CRejectionSamplingRangeOnlyLocalization::RS_observationLikelihood ( const CPose2D & x ) [protected]

Returns the NORMALIZED observation likelihood (linear, not exponential!!!) at a given point of the state space (values in the range [0,1]).

virtual double mrpt::bayes::CRejectionSamplingCapable::RS_observationLikelihood ( const TStateSpace & x ) [protected, pure virtual, inherited]

Returns the NORMALIZED observation likelihood (linear, not exponential!!!) at a given point of the state space (values in the range [0,1]).

bool mrpt::slam::CRejectionSamplingRangeOnlyLocalization::setParams	(	const CLandmarksMap &	beaconsMap,
		const CObservationBeaconRanges &	observation,
		float	sigmaRanges,
		const CPose2D &	oldPose,
		float	robot_z = `0`,
		bool	autoCheckAngleRanges = `true`
	)

The parameters used in the generation of random samples:

Parameters:

beaconsMap	The map containing the N beacons (indexed by their "beacon ID"s). Only the mean 3D position of the beacons is used, the covariance is ignored.
observation	An observation with, at least ONE range measurement.
sigmaRanges	The standard deviation of the "range measurement noise".
robot_z	The height of the robot on the floor (default=0). Note that the beacon sensor on the robot may be at a different height, according to data within the observation object.
autoCheckAngleRanges	Whether to make a simple check for potential good angles from the beacons to generate samples (disable to speed-up the preparation vs. making slower the drawn). This method fills out the member "m_dataPerBeacon".

Returns:: true if at least ONE beacon has been successfully loaded, false otherwise. In this case do not call "rejectionSampling" or an exception will be launch, since there is no information to generate samples.