scispace - formally typeset
Search or ask a question
Author

Chan Gook Park

Other affiliations: Systems Research Institute, Samsung
Bio: Chan Gook Park is an academic researcher from Seoul National University. The author has contributed to research in topics: Kalman filter & Extended Kalman filter. The author has an hindex of 20, co-authored 209 publications receiving 1839 citations. Previous affiliations of Chan Gook Park include Systems Research Institute & Samsung.


Papers
More filters
Journal ArticleDOI
TL;DR: In this article, an adaptive two-stage extended Kalman filter (ATEKF) using an adaptive fading EKF has been proposed to solve the problem of unknown bias.
Abstract: The well-known conventional Kalman filter requires an accurate system model and exact stochastic information. But in a number of situations, the system model has an unknown bias, which may degrade the performance of the Kalman filter or may cause the filter to diverge. The effect of the unknown bias may be more pronounced on the extended Kalman filter (EKF), which is a nonlinear filter. The two-stage extended Kalman filter (TEKF) with respect to this problem has been receiving considerable attention for a long time. Recently, the optimal two-stage Kalman filter (TKF) for linear stochastic systems with a constant bias or a random bias has been proposed by several researchers. A TEKF can also be similarly derived as the optimal TKF. In the case of a random bias, the TEKF assumes that the information of a random bias is known. But the information of a random bias is unknown or partially known in general. To solve this problem, this paper proposes an adaptive two-stage extended Kalman filter (ATEKF) using an adaptive fading EKF. To verify the performance of the proposed ATEKF, the ATEKF is applied to the INS-GPS (inertial navigation system-Global Positioning System) loosely coupled system with an unknown fault bias. The proposed ATEKF tracked/estimated the unknown bias effectively although the information about the random bias was unknown.

156 citations

Journal ArticleDOI
TL;DR: A novel visual-inertial navigation algorithm for low-cost and computationally constrained vehicle in global positioning system denied environments is presented by modeling the state space as the matrix Lie group (LG), based on the recent theory of the invariant Kalman filter.
Abstract: In this paper, we present a novel visual-inertial navigation algorithm for low-cost and computationally constrained vehicle in global positioning system denied environments by modeling the state space as the matrix Lie group (LG), based on the recent theory of the invariant Kalman filter. The multistate constraint Kalman filter (MSCKF) is a well-known visual-inertial odometry algorithm that performs the fusion of the visual and inertial information by constraining each other through the stochastically cloned pose within a sliding window. However, conventional MSCKF (MSCKF-Conv) suffers from the inconsistent state estimates caused by the spurious gain along the unobservable directions, resulting in large estimation errors. To tackle this problem, we extend the concepts of the state and noise of the MSCKF from Euclidean space to matrix LG. We model the state of the MSCKF as the element of the specially customized matrix LG and use the noise uncertainty modeling with the corresponding Lie algebra. The detailed derivation and observability analysis of the proposed filter are provided to prove that the proposed filter is more consistent than the MSCKF-Conv. The proposed MSCKF on matrix LG naturally enforces the state vector to exist in the state space that maintains the unobservability characteristics without any artificial remedies. The performance of the proposed filter is validated through the Monte-Carlo simulation and the real-world experimental dataset.

85 citations

Journal ArticleDOI
TL;DR: Two adaptive filters, such as an adaptive fading Kalman filter (AFKF) and an adaptive two-stage Kalman filters (ATKF), are proposed, which are designed by using the forgetting factor obtained from the innovation information and the stability of the AFKF is analysed.
Abstract: The well-known conventional Kalman filter gives the optimal solution but requires an accurate system model and exact stochastic information. In a number of practical situations, the system model has unknown bias and the Kalman filter with unknown bias may be degraded or even diverged. The two-stage Kalman filter (TKF) to consider this problem has been receiving considerable attention for a long time. Until now, the optimal TKF for system with a constant bias or a random bias has been proposed by several researchers. In case of a random bias, the optimal TKF assumes that the information of a random bias is known. But the information of a random bias is unknown or incorrect in general. To solve this problem, this paper proposes two adaptive filters, such as an adaptive fading Kalman filter (AFKF) and an adaptive two-stage Kalman filter (ATKF). Firstly, the AFKF is designed by using the forgetting factor obtained from the innovation information and the stability of the AFKF is analysed. Secondly, the ATKF to estimate unknown random bias is designed by using the AFKF and the performance of the ATKF is verified by simulation. Copyright © 2006 John Wiley & Sons, Ltd.

83 citations

Journal Article
TL;DR: The compensation method for GPS and an odometers is introduced and new compensation methods are proposed for an odometer to consider the effect of coordinate transformation errors and the scale factor error.
Abstract: For more accurate navigation, lever arm compensation is considered. The compensation method for GPS and an odometer is introduced and new compensation methods are proposed for an odometer to consider the effect of coordinate transformation errors and the scale factor error. The methods are applied to a GPS/INS/odometer integrated system and the simulation and experimental results show its effectiveness. Navigation is defined as all the related theories and technologies for obtaining position, velocity and attitude of a vehicle. With the use of navigation technology, one can know his/her position and can plan trajectory for their destination. Nowadays, especially, the need for the navigation of land vehicles has rapidly increased. In the near future, ubiquitous personal navigation will be spread and navigation will become a more essential technology. The Global Positioning System (GPS) is a satellite- based radio navigation system (1). It allows a user with a receiver to obtain accurate position information anywhere on the globe. It provides position information whose errors would not increase with respect to time. However, a signal from the GPS satellite cannot often arrive at a receiver in an urban area. In those cases, it cannot provide any position information.

74 citations

Journal ArticleDOI
TL;DR: The proposed robust filter is constructed based on the linear approximation methods for a general nonlinear uncertain system with an integral quadratic constraint and derives the important characteristic of the proposed filter, a modified Hinfin performance index.
Abstract: In this correspondence, a robust filter is proposed to effectively estimate the system states in the case where system model uncertainties as well as disturbances are present. The proposed robust filter is constructed based on the linear approximation methods for a general nonlinear uncertain system with an integral quadratic constraint. We also derive the important characteristic of the proposed filter, a modified Hinfin performance index. Analysis results show that the proposed filter has robustness against disturbances such as process and measurement noises, and against parameter uncertainties. Simulation results show that the proposed filter provides a performance improvement

65 citations


Cited by
More filters
Christopher M. Bishop1
01 Jan 2006
TL;DR: Probability distributions of linear models for regression and classification are given in this article, along with a discussion of combining models and combining models in the context of machine learning and classification.
Abstract: Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

10,141 citations

01 Jan 2015
TL;DR: This compact, informal introduction for graduate students and advanced undergraduates presents the current state-of-the-art filtering and smoothing methods in a unified Bayesian framework and learns what non-linear Kalman filters and particle filters are, how they are related, and their relative advantages and disadvantages.
Abstract: Filtering and smoothing methods are used to produce an accurate estimate of the state of a time-varying system based on multiple observational inputs (data). Interest in these methods has exploded in recent years, with numerous applications emerging in fields such as navigation, aerospace engineering, telecommunications, and medicine. This compact, informal introduction for graduate students and advanced undergraduates presents the current state-of-the-art filtering and smoothing methods in a unified Bayesian framework. Readers learn what non-linear Kalman filters and particle filters are, how they are related, and their relative advantages and disadvantages. They also discover how state-of-the-art Bayesian parameter estimation methods can be combined with state-of-the-art filtering and smoothing algorithms. The book’s practical and algorithmic approach assumes only modest mathematical prerequisites. Examples include MATLAB computations, and the numerous end-of-chapter exercises include computational assignments. MATLAB/GNU Octave source code is available for download at www.cambridge.org/sarkka, promoting hands-on work with the methods.

1,102 citations

Journal ArticleDOI
22 Feb 2017-Sensors
TL;DR: A novel method named Deep Convolutional Neural Networks with Wide First-layer Kernels (WDCNN), which can not only achieve 100% classification accuracy on normal signals, but also outperform the state-of-the-art DNN model which is based on frequency features under different working load and noisy environment conditions.
Abstract: Intelligent fault diagnosis techniques have replaced time-consuming and unreliable human analysis, increasing the efficiency of fault diagnosis. Deep learning models can improve the accuracy of intelligent fault diagnosis with the help of their multilayer nonlinear mapping ability. This paper proposes a novel method named Deep Convolutional Neural Networks with Wide First-layer Kernels (WDCNN). The proposed method uses raw vibration signals as input (data augmentation is used to generate more inputs), and uses the wide kernels in the first convolutional layer for extracting features and suppressing high frequency noise. Small convolutional kernels in the preceding layers are used for multilayer nonlinear mapping. AdaBN is implemented to improve the domain adaptation ability of the model. The proposed model addresses the problem that currently, the accuracy of CNN applied to fault diagnosis is not very high. WDCNN can not only achieve 100% classification accuracy on normal signals, but also outperform the state-of-the-art DNN model which is based on frequency features under different working load and noisy environment conditions.

876 citations

Book ChapterDOI
Roy M. Howard1
01 Jan 2002
TL;DR: Chapter 8 establishes the relationship between the input and output power spectral densities of a linear system and the theory is extended to multiple input-multiple output systems.
Abstract: Chapter 8 establishes the relationship between the input and output power spectral densities of a linear system. Limitations on results are carefully detailed and the case of oscillator noise is considered. The theory is extended to multiple input-multiple output systems.

789 citations

Journal ArticleDOI
TL;DR: A survey of the information sources and information fusion technologies used in current in-car navigation systems is presented and the pros and cons of the four commonly used information sources are described.
Abstract: In-car positioning and navigation has been a killer application for Global Positioning System (GPS) receivers, and a variety of electronics for consumers and professionals have been launched on a large scale. Positioning technologies based on stand-alone GPS receivers are vulnerable and, thus, have to be supported by additional information sources to obtain the desired accuracy, integrity, availability, and continuity of service. A survey of the information sources and information fusion technologies used in current in-car navigation systems is presented. The pros and cons of the four commonly used information sources, namely, 1) receivers for radio-based positioning using satellites, 2) vehicle motion sensors, 3) vehicle models, and 4) digital map information, are described. Common filters to combine the information from the various sources are discussed. The expansion of the number of satellites and the number of satellite systems, with their usage of available radio spectrum, is an enabler for further development, in combination with the rapid development of microelectromechanical inertial sensors and refined digital maps.

524 citations