kalmanFilter

Executive Summary

This virtual class provides an interface for all Kalman Filters. Modules which inherit from it will benefit from having the core logic of a sequential filter

Virtual and Private method descriptors

The following table lists all the class methods and their function

Interface methods which remain private

Method Name	Method Function	Protected or Private	Virtual or not
updateState	Loop through measurements and update filter states according to time	public	final
timeUpdate	filter time update	protected	virtual
measurementUpdate	filter measurement update in the presence of measurements	protected	virtual
computeResiduals	compute filter post fit residuals	protected	virtual
customReset	perform addition reset duties in child class	protected	virtual, optional
readFilterMeasurements	read the specific measurements in a child class	protected	virtual, needs to be populated
writeOutputMessages	write the specific measurements in a child class	protected	virtual, needs to be populated
orderMeasurementsChronologically	Runge-Kutta 4 integrator if needed for propagation	private	non-virtual

Module assumptions and limitations

The module inherits all assumptions made while implementing a Kalman filter:

• Observability considerations

• Gaussian covariances

• Linearity limits

• and more

Otherwise, the limitations are governed by the measurement model and dynamics models relative to the required performance.

User Guide

This section lists all the setters and getters that are defined by the interface

Interface methods which remain private

Method Name	Method Function	Necessity
set/getInitialPosition	filter initial position component of the states	optional
set/getInitialVelocity	filter initial velocity component of the states	optional
set/getInitialAcceleration	filter initial acceleration component of the states	optional
set/getInitialBias	filter initial bias component of the states	optional
set/getInitialConsider	filter initial consider component of the states	optional
set/getInitialCovariance	filter initial covariance (square matrix size of state)	necessary
set/getProcessNoise	filter process noise value (square matrix size of state)	necessary
set/getUnitConversionFromSItoState	conversion from SI to internal filter units	optional
set/getMeasurementNoiseScale	filter measurement noise scale factor	optional
setFilterDynamics	provide the dynamics class to the filter	necessary

class KalmanFilter : public SysModel

Kalman Filter interface.

Subclassed by EkfInterface, SRukfInterface

Public Functions

void updateState(uint64_t currentSimNanos) final

Take the relative position measurements and output an estimate of the spacecraft states in the inertial frame.

Parameters:

currentSimNanos – The clock time at which the function was called (nanoseconds)

Returns:

void

void setInitialPosition(const Eigen::VectorXd &initialPositionInput)

Set the filter initial state position vector

Parameters:

Eigen::VectorXd – initial position vector

Returns:

void

std::optional<Eigen::VectorXd> getInitialPosition() const

Get the filter initial state position vector

Returns:

std::optional<Eigen::VectorXd> initial position vector

void setInitialVelocity(const Eigen::VectorXd &initialVelocityInput)

Set the filter initial state velocity vector

Parameters:

Eigen::VectorXd – initial velocity vector

Returns:

void

std::optional<Eigen::VectorXd> getInitialVelocity() const

Get the filter initial state velocity vector

Returns:

std::optional<Eigen::VectorXd> initial velocity vector

void setInitialAcceleration(const Eigen::VectorXd &initialAccelerationInput)

Set the filter initial state acceleration vector

Parameters:

Eigen::VectorXd – initial acceleration vector

Returns:

void

std::optional<Eigen::VectorXd> getInitialAcceleration() const

Get the filter initial state acceleration vector

Returns:

std::optional<Eigen::VectorXd> initial acceleration vector

void setInitialBias(const Eigen::VectorXd &initialBiasInput)

Set the filter initial state bias vector

Parameters:

Eigen::VectorXd – initial bias vector

Returns:

void

std::optional<Eigen::VectorXd> getInitialBias() const

Get the filter initial state bias vector

Returns:

std::optional<Eigen::VectorXd> initial bias vector

void setInitialConsiderParameters(const Eigen::VectorXd &initialConsiderInput)

Set the filter initial state consider parameter vector

Parameters:

Eigen::VectorXd – initial consider parameter vector

Returns:

void

std::optional<Eigen::VectorXd> getInitialConsiderParameters() const

Get the filter initial state consider parameter vector

Returns:

std::optional<Eigen::VectorXd> initial consider parameter vector

void setInitialCovariance(const Eigen::MatrixXd &initialCovariance)

Set the filter initial state covariance

Parameters:

Eigen::MatrixXd – initialCovarianceInput

Returns:

void

Eigen::MatrixXd getInitialCovariance() const

Get the filter initial state covariance

Returns:

Eigen::MatrixXd covarInitial

void setProcessNoise(const Eigen::MatrixXd &processNoise)

Set the filter process noise

Parameters:

Eigen::MatrixXd – processNoiseInput

Returns:

void

Eigen::MatrixXd getProcessNoise() const

Get the filter process noise

Returns:

Eigen::MatrixXd processNoise

void setUnitConversionFromSItoState(double conversion)

Set a unit conversion factor, for instance if desirable to solve for a state in km in the filter, but Basilisk’s outside facing interface is in SI

Parameters:

double – conversion

double getUnitConversionFromSItoState() const

Get a unit conversion factor, for instance if desirable to solve for a state in km in the filter, but Basilisk’s outside facing interface is in SI

Returns:

double unitConversion

void setMeasurementNoiseScale(double measurementNoiseScale)

Set the filter measurement noise scale factor if desirable

Parameters:

double – measurementNoiseScale

Returns:

void

double getMeasurementNoiseScale() const

Get the filter measurement noise scale factor

Returns:

double measurementNoiseScale

void setFilterDynamics(const std::function<const FilterStateVector(double, const FilterStateVector&)> &dynamicsPropagator)

Set the filter equations of motion of the state (X_dot = f(t, X))

Parameters:

Eigen::VectorXd – initialStateInput

Returns:

void