Unscented Kalman filter class. More...
#include <ukf.h>
Public Member Functions | |
| void | correct (const Measurement &measurement) |
| Carries out the correct step in the predict/update cycle. | |
| void | predict (const double referenceTime, const double delta) |
| Carries out the predict step in the predict/update cycle. | |
| Ukf (std::vector< double > args) | |
| Constructor for the Ukf class. | |
| ~Ukf () | |
| Destructor for the Ukf class. | |
Protected Attributes | |
| std::vector< double > | covarWeights_ |
| The weights associated with each sigma point when calculating a predicted estimateErrorCovariance_. | |
| double | lambda_ |
| Used in weight generation for the sigma points. | |
| std::vector< Eigen::VectorXd > | sigmaPoints_ |
| The UKF sigma points. | |
| std::vector< double > | stateWeights_ |
| The weights associated with each sigma point when generating a new state. | |
| bool | uncorrected_ |
| Used to determine if we need to re-compute the sigma points when carrying out multiple corrections. | |
| Eigen::MatrixXd | weightedCovarSqrt_ |
| This matrix is used to generate the sigmaPoints_. | |
Detailed Description
Unscented Kalman filter class.
Implementation of an unscenter Kalman filter (UKF). This class derives from FilterBase and overrides the predict() and correct() methods in keeping with the discrete time UKF algorithm. The algorithm was derived from the UKF Wikipedia article at (http://en.wikipedia.org/wiki/Kalman_filter#Unscented_Kalman_filter) ...and this paper: J. J. LaViola, Jr., “A comparison of unscented and extended Kalman filtering for estimating quaternion motion,” in Proc. American Control Conf., Denver, CO, June 4–6, 2003, pp. 2435–2440 Obtained here: http://www.cs.ucf.edu/~jjl/pubs/laviola_acc2003.pdf
Constructor & Destructor Documentation
| RobotLocalization::Ukf::Ukf | ( | std::vector< double > | args | ) | [explicit] |
Member Function Documentation
| void RobotLocalization::Ukf::correct | ( | const Measurement & | measurement | ) | [virtual] |
Carries out the correct step in the predict/update cycle.
- Parameters:
-
[in] measurement - The measurement to fuse with our estimate
Implements RobotLocalization::FilterBase.
| void RobotLocalization::Ukf::predict | ( | const double | referenceTime, |
| const double | delta | ||
| ) | [virtual] |
Carries out the predict step in the predict/update cycle.
Projects the state and error matrices forward using a model of the vehicle's motion.
- Parameters:
-
[in] referenceTime - The time at which the prediction is being made [in] delta - The time step over which to predict.
Implements RobotLocalization::FilterBase.
Member Data Documentation
std::vector<double> RobotLocalization::Ukf::covarWeights_ [protected] |
double RobotLocalization::Ukf::lambda_ [protected] |
std::vector<Eigen::VectorXd> RobotLocalization::Ukf::sigmaPoints_ [protected] |
std::vector<double> RobotLocalization::Ukf::stateWeights_ [protected] |
bool RobotLocalization::Ukf::uncorrected_ [protected] |
Eigen::MatrixXd RobotLocalization::Ukf::weightedCovarSqrt_ [protected] |
The documentation for this class was generated from the following files: