• Journal of the Korea Institute of Military Science and Technology
• /
• v.9 no.3
• /
• pp.132-140
• /
• 2006

In this paper, a new fault detection method for the extended Kalman filter, which uses a N-step predictor, is proposed. The N-step predictor performs the only time propagations for N-step intervals without measurement updates and its output is used as a monitoring signal for the fault detection. A consistency between the extended Kalman filter and the N-step predictor is tested to detect a fault. A test statistic is defined by the difference between the extended Kalman filter and the N-step predictor. The proposed method is applied to strapdown inertial navigation system (SDINS). By computer simulation, it is shown that the proposed method detects a fault effectively.

• Journal of Positioning, Navigation, and Timing
• /
• v.8 no.3
• /
• pp.107-117
• /
• 2019

When a vehicle moves with a high rotation rate, it is not easy to measure the angular velocity using an off-the-shelf gyroscope. If the angular velocity is estimated using the extended Kalman filter in the gyro-free inertial navigation system, the effect of the accelerometer error and initial angular velocity error can be reduced. In this paper, in order to improve the navigation performance of the gyro-free inertial navigation system, an angular velocity estimation method is proposed based on an extended Kalman filter with an accelerometer random bias error model. In order to show the validity of the proposed estimation method, angular velocities and navigation outputs of a vehicle with 3 rev/s rotation rate are estimated. The results are compared with estimates by other methods such as the integration and an extended Kalman filter without an accelerometer random bias error model. The proposed method gives better estimation results than other methods.

• 제어로봇시스템학회:학술대회논문집
• /
• 1996.10b
• /
• pp.860-863
• /
• 1996

In this paper, we formulate the extended Kalman filter for calibration of gimballed inertial navigation system (GINS) at a pure navigation mode with 1500 ft/sec initial velocity and compare its performance to the linear Kalman filter's by using Monte-Carlo analysis method. It has been shown that estimation performance of the extended Kalman filter is better than that of the linear Kalman filter.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.22 no.2
• /
• pp.115-123
• /
• 2018

This paper discusses extended Kalman filter method for solving learning problems of multilayered neural networks. A lot of learning algorithms for deep layered network are sincerely suffered from complex computation and slow convergence because of a very large number of free parameters. We consider an efficient learning algorithm for deep neural network. Extended Kalman filter method is applied to parameter estimation of neural network to improve convergence and computation complexity. We discuss how an efficient algorithm should be developed for neural network learning by using Extended Kalman filter.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.37 no.9
• /
• pp.636-645
• /
• 1988

A new filter algorithm is suggested improving structurally the conventional extended Kalman filter of which the performance is dependent on the selection of the reference axes, by use of line-of-sight axes and gain rotation technique. The implementation method using microcomputer which implements tracking Kalman filter is introduced in terms of hardware and software. Then, through the simulation the performance of suggested filter is compared with that of conventional extended Kalman filter and the possibility of the real time tracking of moving target is investigated.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.978-983
• /
• 2003

In this paper, we propose a fault detection method for extended Kalman filter in decentralized filter structure. To detect a fault, a consistency between filter output and a monitoring signal is tested. State propagators are used to obtain the monitoring signal. However, the output of state propagator increases in magnitude and finally diverges as time runs. To solve such problem, two-propagator method was proposed for linear system. Two propagators are reset by Kalman filter output, alternatively, to avoid divergence. But a test statistics change abruptly at the reset instant in that method. Hence a N-step propagator method is proposed to fix up the problem. In the N-step propagator, only time propagations are performed from k-N+1 step to k step without measurement updates. A test statistics are defined by errors and its covariance between extended Kalman filter and N-step propagator. These fault detection methods are applied to integrated strapdown inertial navigation system (SDINS). By computer simulation, it is shown that the proposed methods detect a fault effectively.

• Proceedings of the KIEE Conference
• /
• 2004.11c
• /
• pp.349-351
• /
• 2004

The classical dynamic backpropagation learning algorithm has the problems of learning speed and the determine of learning parameter. The Extend Kalman Filter(EKF) is used effectively for a state estimation method for a non linear dynamic system. This paper presents a learning algorithm using Dual Extended Kalman Filter(DEKF) for Fully Recurrent Neural Network(FRNN). This DEKF learning algorithm gives the minimum variance estimate of the weights and the hidden outputs. The proposed DEKF learning algorithm is applied to the system identification of a nonlinear SISO system and compared with dynamic backpropagation learning algorithm.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.63 no.2
• /
• pp.277-283
• /
• 2014

This paper proposes robot localization using Fuzzy-Extended Kalman Filter algorithm of the mobile robots equipped with least sensors. In order to improve the accuracy of the localization, we usually add the sensors or equipment. However, it increases the simulation time and expenses. This paper solves this problem using only the odometer and ultrasonic sensors to get the localization with the Fuzzy-Extended Kalman Filter algorithm method. By inputting the robot's angular velocity, sensor data variation, and residual errors into the fuzzy algorithm, we get the sensor weight factor to decide the sensor's importance. The performance of the designed method shows by the simulation and Pioneer 3-DX mobile robot test in the indoor environment.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.36 no.12
• /
• pp.908-917
• /
• 1987

Most of moving targets are modelled as nonlinear dynamic equations. In recent years, the extended Kalman filter is frequently used for estimating their behaviors. The conditional Gaussian filter is more suitable than extended kalman filter in the filtering problem of nonlinear systems. But extended Kalman filter and conditional Gaussian filter often do not give optimal estimates and fail to track target trajectories because of its properties. Therefore it is desirable to use adaptive techniques to adapt target maneuvers. In this paper, we will discuss adaptive filtering technique using innovation process based on extended Kalman filter in real time, and suggest another maneuver estimation method using MRAS technique.

• Smart Media Journal
• /
• v.13 no.5
• /
• pp.9-18
• /
• 2024

Multi-Object trajectory modeling is a major challenge in MOT. CenterTrack tried to solve this problem with a Heatmap-based method that tracks the object center position. However, it showed limited performance when tracking objects with complex movements and nonlinearities. Considering the degradation factor of CenterTrack as the dynamic movement of pedestrians, we integrated the EKF into CenterTrack. To demonstrate the superiority of our proposed method, we applied the existing KF and UKF to CenterTrack and compared and evaluated it on various datasets. The experimental results confirmed that when EKF was integrated into CenterTrack, it achieved 73.7% MOTA, making it the most suitable filter for CenterTrack.