UKF Class Reference

Inheritance diagram for UKF:

[legend]

Collaboration diagram for UKF:

[legend]

List of all members.

Public Member Functions
	UKF (int numStates, int numObservations, double k=0, double alpha=1e-3, double beta=2)
virtual Image< double >	getNextState (const Image< double > &X, int k)=0
	The function to move from one state to another (need to be implemented).
virtual Image< double >	getObservation (const Image< double > &X, int k)=0
	The function to predict the observation from the current state.
void	predictState (const Image< double > &noise=Image< double >())
	Predict the next state and covariance.
void	predictObservation (const Image< double > &noise=Image< double >())
	Predict the observations.
void	update (const Image< double > &z, const Image< double > &noise)
	Update the state and covariance given the observation z.
double	getLikelihood (const Image< double > &z, const Image< double > &observationNoise)
	Get the likelihood of a mesurment.
Protected Member Functions
Image< double >	getSigmaLocations (const Image< double > &state, const Image< double > &sigma, double gamma)
Protected Attributes
int	itsNumStates
int	itsNumObservations
double	itsAlpha
double	itsK
double	itsBeta
Image< double >	itsState
Image< double >	itsSigma
Image< double >	itsR
Image< double >	itsQ
Image< double >	itsSigmaLocations
Image< double >	itsNewStates
Image< double >	itsNewZ
Image< double >	itsPredictedZ
Image< double >	itsPredictedZSigma

---

Detailed Description

Definition at line 48 of file UKF.H.

---

Member Function Documentation

double UKF::getLikelihood	(	const Image< double > &	z,
		const Image< double > &	observationNoise
	)

Get the likelihood of a mesurment.

Definition at line 170 of file UKF.C.

References exp(), Image< T >::getVal(), Image< T >::initialized(), matrixInv(), matrixMult(), sqrt(), and transpose().

virtual Image<double> UKF::getNextState	(	const Image< double > &	X,
		int	k
	)			[pure virtual]

The function to move from one state to another (need to be implemented).

Implemented in EyeFilter, and ParticleTracker.

Referenced by predictState().

virtual Image<double> UKF::getObservation	(	const Image< double > &	X,
		int	k
	)			[pure virtual]

The function to predict the observation from the current state.

Implemented in EyeFilter, and ParticleTracker.

Referenced by predictObservation().

void UKF::predictObservation

(

const Image< double > &

noise = Image<double>()

)

Predict the observations.

Definition at line 128 of file UKF.C.

References ASSERT, getObservation(), Image< T >::getVal(), Image< T >::getWidth(), Image< T >::initialized(), matrixMult(), Image< T >::setVal(), Image< T >::size(), transpose(), and ZEROS.

void UKF::predictState

(

const Image< double > &

noise = Image<double>()

)

Predict the next state and covariance.

Definition at line 83 of file UKF.C.

References ASSERT, Image< T >::getDims(), getNextState(), Image< T >::getVal(), Image< T >::getWidth(), Image< T >::initialized(), matrixMult(), Image< T >::setVal(), Image< T >::size(), transpose(), and ZEROS.

Referenced by main().

void UKF::update	(	const Image< double > &	z,
		const Image< double > &	noise
	)

Update the state and covariance given the observation z.

Definition at line 204 of file UKF.C.

References Image< T >::getVal(), Image< T >::getWidth(), matrixInv(), matrixMult(), NO_INIT, Image< T >::size(), transpose(), and ZEROS.

Referenced by main().

---

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

• src/BayesFilters/UKF.H
• src/BayesFilters/UKF.C