rtabmap::BayesFilter Class Reference

Recursive Bayesian filter for loop-closure hypothesis estimation in RTAB-Map. More...

#include <BayesFilter.h>

Public Member Functions
	BayesFilter (const ParametersMap &parameters=ParametersMap())
	Constructs a Bayes filter with default or custom parameters.
virtual void	parseParameters (const ParametersMap &parameters)
	Updates internal settings from the parameter map.
bool	computePosterior (const Memory *memory, const std::map< int, float > &likelihood)
	Runs one Bayes filter iteration (prediction + update).
void	reset ()
	Clears posterior, prediction matrix and cached neighbor indices.
void	setPredictionLC (const std::string &prediction)
	Sets the loop-closure prediction model from a space-separated string.
const std::vector< int > &	getPosteriorIds () const
	The locations the posterior is over, ascending by id.
const std::vector< float > &	getPosteriorValues () const
	The probability of each location of getPosteriorIds(), in the same order.
float	getVirtualPlacePrior () const
	Returns the virtual place prior threshold.
bool	isPredictionSparse () const
	Whether the prediction is being kept in its sparse form rather than as a matrix.
const std::vector< double > &	getPredictionLC () const
	Returns the loop-closure prediction model as a vector of values.
std::string	getPredictionLCStr () const
	Returns the loop-closure prediction model as a space-separated string.
cv::Mat	generatePrediction (const Memory *memory, const std::vector< int > &ids)
	Builds or updates the prediction (transition) matrix for the given signature ids.
unsigned long	getMemoryUsed () const
	Estimates memory usage of this object and its internal containers.

Detailed Description

Recursive Bayesian filter for loop-closure hypothesis estimation in RTAB-Map.

This class implements the prediction and update steps of a Bayes filter used to estimate the posterior probability over candidate locations (signatures) in working memory. It is typically called by Rtabmap after likelihood values have been computed from visual word comparisons.

The filter operates in two steps on each iteration:

• Prediction: builds a transition matrix from the memory graph and multiplies it with the previous posterior to obtain the prior.
• Update: multiplies the prior by the observation likelihood and normalizes the result.

The prediction matrix is built from neighbor relationships in Memory, using a Gaussian-like model configured through Parameters::kBayesPredictionLC(). A virtual place (negative signature id, see Memory::kIdVirtual) represents the hypothesis that the current observation comes from a new location.

Related parameters (see Parameters):

• Parameters::kBayesPredictionLC() — transition probabilities per graph depth level.
• Parameters::kBayesVirtualPlacePriorThr() — prior for the virtual place.
• Parameters::kBayesFullPredictionUpdate() — regenerate the full prediction matrix each iteration.
• Parameters::kBayesSparsePrediction() — keep the prediction sparse and multiply it sparsely.

See also

Memory::getNeighborsId()

Rtabmap

Definition at line 81 of file BayesFilter.h.

Constructor & Destructor Documentation

BayesFilter()

rtabmap::BayesFilter::BayesFilter

(

const ParametersMap &

parameters = ParametersMap()

)

Constructs a Bayes filter with default or custom parameters.

Parameters

parameters

Optional parameter map (Bayes group keys). Defaults are used for missing keys.

Member Function Documentation

parseParameters()

virtual void rtabmap::BayesFilter::parseParameters ( const ParametersMap &  parameters )

virtual

Updates internal settings from the parameter map.

Parameters

parameters

Map containing Bayes group keys.

computePosterior()

bool rtabmap::BayesFilter::computePosterior	(	const Memory *	memory,
		const std::map< int, float > &	likelihood
	)

Runs one Bayes filter iteration (prediction + update).

Given a likelihood map over signature ids, computes and stores the normalized posterior. The prediction matrix is generated or updated from Memory using the ids present in likelihood.

Read the result with getPosteriorIds() and getPosteriorValues().

Parameters

memory	Working memory instance (must not be null).
likelihood	Observation likelihood per signature id (must not be empty).

Returns

False on error (null memory, empty likelihood, or invalid prediction model), the posterior being left unchanged.

setPredictionLC()

void rtabmap::BayesFilter::setPredictionLC

(

const std::string &

prediction

)

Sets the loop-closure prediction model from a space-separated string.

Format: {Vp, Lc, l1, l2, l3, ...} where:

• Vp — virtual place probability. This is the probability to move to a new place (unvisited location).
• Lc — loop closure (depth 0) probability. This is the probability to stay at the same location.
• l1, l2, ... — probabilities for neighbors at increasing graph depth levels. This is the probability to move to a neighbor at the given depth level.

Each value must be in [0, 1]. At least two values are required. Invalid strings are rejected and the previous model is kept.

Parameters

prediction

Space-separated list of probabilities (same format as Parameters::kBayesPredictionLC()).

getPosteriorIds()

const std::vector< int > & rtabmap::BayesFilter::getPosteriorIds ( ) const

inline

The locations the posterior is over, ascending by id.

The virtual place (Memory::kIdVirtual) is the first of them when it is one.

Definition at line 138 of file BayesFilter.h.

getPosteriorValues()

const std::vector< float > & rtabmap::BayesFilter::getPosteriorValues ( ) const

inline

The probability of each location of getPosteriorIds(), in the same order.

Definition at line 143 of file BayesFilter.h.

getVirtualPlacePrior()

float rtabmap::BayesFilter::getVirtualPlacePrior

(

)

const

Returns the virtual place prior threshold.

Returns

Value in [0, 1] used when building the virtual place row of the prediction matrix.

isPredictionSparse()

bool rtabmap::BayesFilter::isPredictionSparse

(

)

const

Whether the prediction is being kept in its sparse form rather than as a matrix.

False when Parameters::kBayesSparsePrediction() is disabled, and over a model whose values sum to less than 1, which leaves no zero in a column to keep out of the values.

getPredictionLC()

const std::vector< double > & rtabmap::BayesFilter::getPredictionLC

(

)

const

Returns the loop-closure prediction model as a vector of values.

Returns

Vector in the format {Vp, Lc, l1, l2, l3, ...}.

getPredictionLCStr()

std::string rtabmap::BayesFilter::getPredictionLCStr

(

)

const

Returns the loop-closure prediction model as a space-separated string.

Returns

String representation of getPredictionLC().

generatePrediction()

cv::Mat rtabmap::BayesFilter::generatePrediction	(	const Memory *	memory,
		const std::vector< int > &	ids
	)

Builds or updates the prediction (transition) matrix for the given signature ids.

Rows and columns correspond to ids. Neighbor links are queried from Memory to fill transition probabilities according to getPredictionLC(). When ids match the current posterior keys, the cached matrix may be returned without recomputation.

When the prediction is being kept sparse, the matrix is expanded from it rather than kept: it costs the memory that keeping the prediction sparse is saving, so ask for it to read, dump or compare the prediction, not on every iteration.

Parameters

memory	Working memory instance (must not be null).
ids	Ordered list of signature ids (often includes Memory::kIdVirtual as first element).

Returns

Square CV_32FC1 matrix of size ids.size() × ids.size().

getMemoryUsed()

unsigned long rtabmap::BayesFilter::getMemoryUsed

(

)

const

Estimates memory usage of this object and its internal containers.

Returns

Approximate memory footprint in bytes.

---

The documentation for this class was generated from the following file:

• corelib/include/rtabmap/core/BayesFilter.h