Halil Ersin Soken

UKF-Based Reconfigurable Attitude Parameters Estimation and Magnetometer Calibration

A reconfigurable unscented Kalman filter (UKF)-based estimation algorithm for magnetometer biases and scale factors is proposed as a part of the attitude estimation scheme of a pico satellite. Unlike existing algorithms, in this paper, scale factors are not treated, together with other parameters, as part of the state vector; they are estimated separately. After satisfying the conditions of the proposed stopping rule for bias estimation, UKF reconfigures itself for estimation of attitude parameters alone.

Robust Kalman filtering for small satellite attitude estimation in the presence of measurement faults

Abstract In normal working conditions it is possible to achieve sufficient attitude estimation accuracy for a satellite using regular Kalman filter algorithm. On the other hand, when there is a fault in the measurements, the Kalman filter fails in providing the required accuracy and may even collapse over time. In this paper, a Robust Kalman filtering method is proposed for the attitude estimation problem. By using the proposed method both the Extended Kalman Filter and Unscented Kalman Filter are modified and the new algorithms, which are robust against measurement malfunctions, are called Robust Extended Kalman Filter and Robust Unscented Kalman Filter, respectively. A multiple scale factor based adaptation scheme is preferred for adapting the filters so only the data of the faulty sensor is scaled and any unnecessary information loss is prevented. The proposed algorithms are demonstrated for attitude estimation of a small satellite and performances of these two robust Kalman filters are compared in case of different measurement faults. The application of the algorithm is discussed for small satellite missions where the attitude accuracy depends on a limited number of measurements.

Robust Estimation of UAV Dynamics in the Presence of Measurement Faults

This study introduces a robust Kalman filter (RKF) with a filter-gain correction for cases of measurement malfunctions. Using defined variables called measurement-noise scale factors, the faulty measurements are taken into consideration with a small weight and the estimations are corrected without affecting the characteristics of the accurate ones. In this study, RKF algorithms with single and multiple scale factors are proposed and applied for the state estimation process of an unmanned aerial vehicle (UAV) platform. The results of these algorithms are compared for different types of measurement faults, and recommendations for their utilization are given.

Adaptive Fading UKF with Q-Adaptation: Application to Picosatellite Attitude Estimation

AbstractThe unscented Kalman filter (UKF) is a filtering algorithm that gives sufficiently good estimation results for estimation problems of nonlinear systems even when high nonlinearity is in question. However, in the case of system uncertainty the UKF becomes inaccurate and diverges in time. In other words, if any change occurs in the process noise covariance, which is known a priori, the filter fails. This study introduces a novel adaptive fading UKF algorithm based on the correction of process noise covariance (Q-adaptation) for the case of mismatches with the model. By the use of a newly proposed adaptation scheme for the conventional UKF algorithm, change in the noise covariance is detected and corrected. Differently from most of the existing adaptive UKF algorithms, covariance is not updated at each step; it has only been corrected when the change in the process noise covariance is detected, and that brings about a noteworthy reduction in the computational burden. The proposed algorithm is tested ...

Reconfigurable UKF for In-Flight Magnetometer Calibration and Attitude Parameter Estimation

Abstract In this study a reconfigurable unscented Kalman filter (UKF) based algorithm for the estimation of magnetometer biases and scale factors is proposed as a part of the attitude estimation scheme of a pico satellite. Algorithm is composed of two stages; in first stage UKF estimates magnetometer biases and scale factors as well as six attitude parameters of the satellite. Differently from the existing algorithms, scale factors are not treated together with the other parameters as a part of the state vector; three scale factors are estimated via a newly proposed extension for the UKF. After a convergence rule for the biases second stage starts and the UKF reconfigures itself for the estimation of only attitude parameters. At this stage filter regards the biases and scale factors estimated at the initial stage. Proposed algorithm is simulated for attitude estimation of a pico satellite which has three magnetometers and three rate gyros as measurement sensors.

In flight magnetometer calibration via unscented Kalman filter

In this study an unscented Kalman filter (UKF) based algorithm for the estimation of magnetometer biases and scale factors is proposed as a part of the attitude estimation scheme of a small satellite. UKF estimates magnetometer biases and scale factors as well as six attitude parameters of the satellite. Differently from the existing algorithms, scale factors are not treated together with the other parameters as a part of the state vector; three scale factors are estimated via a newly proposed extension for the UKF. Proposed algorithm is simulated for attitude estimation of a pico satellite which has three magnetometers and three rate gyros as measurement sensors. Besides, effects of non-calibrated magnetometers on attitude determination are shown via supplementary simulations.

Attitude and attitude rate estimation for a nanosatellite using SVD and UKF

Singular value decomposition (SVD) method and unscented Kalman filter (UKF) are integrated to estimate the attitude and attitude rates of a nanosatellite recursively. First the SVD method minimizes the Wahbas loss function to find the optimal solution for the attitude on the basis of magnetometer and sun sensor vector measurements. Then the UKF uses this attitude information as the measurements for providing more accurate attitude estimates even when the satellite is in eclipse. The “rotation angle error covariance matrix” calculated for the estimations of the SVD method are regarded as the measurement noise covariance for the UKF. Discussions for the UKF tuning are included specifically for the eclipse period where the SVD method fails and practically there is no measurements incoming to the filter.

Adaptive Tuning of the Unscented Kalman Filter for Satellite Attitude Estimation

AbstractDetermining the process noise covariance of the unscented Kalman filter (UKF) is a difficult procedure. The analytical approximation method gives satisfactory results in certain cases, but it fails when generalized for the estimation of the extended states, such as the case that sensor biases or scale factors are included in the state vector. The main aim of this research is to find an appropriate tuning algorithm for the process noise covariance of the UKF when the magnetometer biases are estimated, as well as attitude and gyro biases. In this sense, an adaptive tuning method for an UKF that is used for satellite attitude estimation is given and the adaptive UKF algorithm is tested in various scenarios for the attitude and sensor bias estimation. The given adaptation method is an easy way of tuning the filter, especially in the absence of any analytical approximation for the calculation of the process noise covariance, and the performed simulations show that by using the adaptive UKF, it is possi...

In-orbit estimation of time-varying residual magnetic moment

A method for in-orbit estimation of time-varying residual magnetic moment (RMM) is presented. By use of a simple approach, the covariance of the Kalman filter is adapted to get better tracking in case of unexpected abrupt changes in the RMM without sacrificing estimation accuracy. The proposed method does not need a priori information about the magnitude of the change and assures both accurate estimation and good tracking performance for changes with different magnitudes.

Fault Tolerant Estimation of UAV Dynamics via Robust Adaptive Kalman Filter

A covariance scaling based robust adaptive Kalman filter (RAKF) algorithm is developed for the case of sensor/actuator faults. The proposed RAKF uses variable scale factors for scaling the process and measurement noise covariances and eliminating the effect of the faults on the estimation procedure. At first, the existing covariance estimation based adaptation techniques are reviewed. Then the covariance scaling methods with single and multiple factors are discussed. After choosing the efficient adaptation method an overall concept for the RAKF is proposed. In this concept, the filter initially isolates the fault, either in the sensors or in the actuators, and then it applies the required adaptation process such that the estimation characteristic is not deteriorated. The performance of the proposed filters is investigated via simulations for the UAV state estimation problem. The results of the presented algorithms are compared for different types of sensor/actuator faults and recommendations about their application are given within this scope.