• Transactions of the Korean Society of Automotive Engineers
• /
• v.20 no.6
• /
• pp.126-131
• /
• 2012

Tire force is one of important parameters which determine vehicle dynamics. However, it is hard to measure tire force directly through sensors. Not only the sensor is expensive but also installation of sensors on harsh environments is difficult. Therefore, estimation algorithms based on vehicle dynamic models are introduced to estimate the tire forces indirectly. In this paper, an estimation system for estimating lateral force and states is suggested. The state-space equation is constructed based on the 3-DOF bicycle model. Extended Kalman Filter, Unscented Kalman Filter and Ensemble Kalman Filter are used for estimating states on the nonlinear system. Performance of each algorithm is evaluated in terms of RMSE (Root Mean Square Error) and maximum error.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• v.1
• /
• pp.277-282
• /
• 2006

It is well known that Kalman filtering is an optimal real-time data fusion method for GPS/INS integration. However, it has some limitations in terms of stability, adaptability and observability. A Kalman filter can perform optimally only when its dynamic model is correctly defined and the noise statistics for the measurement and process are completely known. It is found that estimated Kalman filter states could be influenced by several factors, including vehicle dynamic variations, filter tuning results, and environment changes, etc., which are difficult to model. Neural networks can map input-output relationships without apriori knowledge about them; hence a proper designed neural network is capable of learning and extracting these complex relationships with enough training. This paper presents a GPS/INS integrated system that combines Kalman filtering and neural network algorithms to improve navigation solutions during GPS outages. An Extended Kalman filter estimates INS measurement errors, plus position, velocity and attitude errors etc. Kalman filter states, and gives precise navigation solutions while GPS signals are available. At the same time, a multi-layer neural network is trained to map the vehicle dynamics with corresponding Kalman filter states, at the same rate of measurement update. After the output of the neural network meets a similarity threshold, it can be used to correct INS measurements when no GPS measurements are available. Selecting suitable inputs and outputs of the neural network is critical for this hybrid method. Detailed analysis unveils that some Kalman filter states are highly correlated with vehicle dynamic variations. The filter states that heavily impact system navigation solutions are selected as the neural network outputs. The principle of this hybrid method and the neural network design are presented. Field test data are processed to evaluate the performance of the proposed method.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.12 no.4
• /
• pp.223-229
• /
• 2017

An angle error is a factor obstructing to track accurate position in tracking radars. And the noise incurring the angle error can be divided as follows; thermal noise and glint. In general, Extended Kalman filter used in tracking radars is designed with considering thermal noise only. The Extended Klaman filter uses a fixed measurement error covariance when updating an estimate state by using ahead state and measurement. But, a noise power varies according to the range. Therefore we purposes the adaptive Kalman filter which changes the measurement noise covariance according to the range. In this paper, we compare the performance of the Extended Kalman filter and the proposed adaptive Kalman filter by considering KSLV-I (Korean Satellite Launch Vehicles).

• Proceedings of the KIEE Conference
• /
• 2004.05a
• /
• pp.41-43
• /
• 2004

This paper is concerned with attitude estimation using low cost, small-sized accelerometers and gyroscopes. A two step extended Kalman filter is proposed, which adaptively compensates external acceleration. External acceleration is the main source of estimation error. In the proposed filter, direction of external acceleration is estimated. According to the estimated direction, the accelerometer measurement covariance matrix of the two step extended Kalman filter is adjusted. The proposed algorithm is verified through experiments.

• Journal of Aerospace System Engineering
• /
• v.10 no.1
• /
• pp.21-28
• /
• 2016

Orbit determination problems have been interest of many researchers for long time. Due to the high nonlinearity of the equation of motion and the measurement model, it is necessary to linearize the both equations. To avoid linearization, the filter based on Fokker-Planck equation is designed. with the extended Kalman filter update mechanism, in which the associated Fokker-Planck equation was solved efficiently and accurately via discrete quadrature and the measurement update was done through the extended Kalman filter update mechanism. This filter based on the DQMOM and the EKF update is applied to the orbit determination problem with appropriate modification to mitigate the filter smugness. Unlike the extended Kalman filter, the hybrid filter based on the DQMOM and the EKF update does not require the burdensome evaluation of the Jacobian matrix and Gaussian assumption for the system, and can still provide more accurate estimations of the state than those of the extended Kalman filter especially when measurements are sparse. Simulation results indicate that the advantages of the hybrid filter based on the DQMOM and the EKF update make it a promising alternative to the extended Kalman filter for orbit estimation problems.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.22 no.3
• /
• pp.220-227
• /
• 2014

This paper presents about multi-rate sensors' synchronization and filter fusion via a sigmoid function of the Kalman filter. To synchronize multi-rate sensors, the estimation states of the Kalman filter is modified. A specific matrix that makes the filter choose sensor values only updated is multiplied to measurement matrix. For the filter that has weak points on some criteria, filter fusion is suggested by using sigmoid function. Modified kalman filter is tested with practical case. A sigmoid function was designed for the test and the performance of the modified function is estimated with respect to conventional Kalman filter. Unscented Kalman filter is used to the base filter of the suggested filter because of its stability.

• 제어로봇시스템학회:학술대회논문집
• /
• 1987.10b
• /
• pp.464-469
• /
• 1987

The Kalman filter is an optimal linear estimator that has been an active research topic for the past three decades. The scheme has become the milestone of modern filtering, and it is applied to many areas including navigations and controls of free vehicle. The Kalman filter technique is matured. But some problems are still remained to be resolved. The prevention of divergence induced by digital implementation, nonoptimal application for nonlinear system, and application to non-Gaussian processes are some of the problems. This paper surveys the problems. The square root filtering is suggested to prevent the divergence. The extended Kalman filter is used for nonlinear systems. And, many other approaches to Kalman-like optimal estimators are also investigated.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.26 no.1
• /
• pp.10-21
• /
• 2023

For various target tracking applications, it is well known that the Kalman filter is the optimal estimator(in the minimum mean-square sense) to predict and estimate the state(position and/or velocity) of linear dynamical systems driven by Gaussian stochastic noise. In the case of nonlinear systems, Extended Kalman filter(EKF) and/or Unscented Kalman filter(UKF) are widely used, which can be viewed as approximations of the(linear) Kalman filter in the sense of the conditional expectation. However, to implement EKF and UKF, the exact dynamical model information and the statistical information of noise are still required. In this paper, we propose the recurrent neural-network based Kalman filter, where its Kalman gain is obtained via the proposed GRU-LSTM based neural-network framework that does not need the precise model information as well as the noise covariance information. By the proposed neural-network based Kalman filter, the state estimation performance is enhanced in terms of the tracking error, which is verified through various linear and nonlinear tracking problems with incomplete model and statistical covariance information.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.16 no.28
• /
• pp.31-37
• /
• 1993

Box-Jenkins models have some important limitations to the procedure : (a) They require a great deal of time, efforts and expertise for the model identification. (b) They require an extensive amount of past observations to identify an acceptable model. (c) The model selected is a constant model in time. Therefore, the Kalman Filter is recommended as a technique to overcome the three problems mentioned above. The research reported here uses the Kalman Filter algorithm to propose Kalman-AR(p) model. The data analysis shows that the Kalman-AR(p) model proposed can be used to resolve the problems of Box-Jenkins AR(p)model. It is seen that the Kalman Filter has great potentials for real-time industrial applications.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.561-564
• /
• 2004

This paper proposes a new application of Kalman filter to estimate speed sensorless DC motor. Kalman filter can estimate the system state variables accurately; even the system input is disturbed with noise. In the design, the mathematical model of DC motor in discrete state-space form will be created; the speed of DC motor which is considered as state variable and can be estimated by using Kalman filter. In the experiment; TMS320C31 floating point digital signal processor is used for hardware implementation, the input is disturbed with/and without white noise in the experiment. The experimental results show the speed of DC motor which is estimated by Kalman filter has good accuracy when compared with the results from tacho-meter.