The quality and precision of tracking manuevering targets under large clutter is highly dependent on both the data association and state estimation algorithms. In this study, measurement-to track association problem w...
详细信息
The quality and precision of tracking manuevering targets under large clutter is highly dependent on both the data association and state estimation algorithms. In this study, measurement-to track association problem was discussed and the optimal association problem was shown to be a markov Decision Process. The problemmodel considers the batch measurements in a time interval. The optimization problem has an heavy computational load, therefore the rollout algorithm is used to solve this problem. The approximate solution to the association problem is a new approach and it does not exist in the literature. The algorithm was applied to a tracking scenario and its efficiency is demonstrated in the simulations part.
In this paper, two proposed Track Before Detect (TBD) algorithms for spawning targets on the basis of raw radar measurements are described. These algorithms are developed by using multi-model particle filter method. T...
详细信息
In this paper, two proposed Track Before Detect (TBD) algorithms for spawning targets on the basis of raw radar measurements are described. These algorithms are developed by using multi-model particle filter method. To improve the efficiency of particle filter a novel reduced order model is introduced. The algorithms are confirmed by using the simulation results and their performances are analyzed on the basis of the probability of target existence and Root mean Square (RmS) estimation accuracy for very low Signal-to-Noise Ratio (SNR) targets.
In this study Imm-PF (Interacting multiple model - Particle Filter) algorithm has been developed in order to achieve increased performance for tracking maneuvering targets. Track lost rate and gate sizes are determine...
详细信息
In this study Imm-PF (Interacting multiple model - Particle Filter) algorithm has been developed in order to achieve increased performance for tracking maneuvering targets. Track lost rate and gate sizes are determined as the performance criteria. The developed algorithm has been compared with standard Imm algorithm. Effects of the number of particles on the performance is analysed.
In this paper, estimating the states of an object extension characterized by a collectively moving ballistic object group (cluster) problem is analyzed. Recently, a Bayesian approach to extended object tracking using ...
详细信息
In this paper, estimating the states of an object extension characterized by a collectively moving ballistic object group (cluster) problem is analyzed. Recently, a Bayesian approach to extended object tracking using randommatrices has been proposed. This new approach simultaneously estimates the ellipsoidal shape together with the kinematics of a group of ballistic targets. Target group that is tracked consists of subsequent projectiles. JPDA framework is used together with this new approach to emphasize the pros and cons of both approaches.
In this work, a mathematical model for stochastic uncertain systems where the system uncertainty is handled by polynomial chaos method is developed. For uncertain systems where the system uncertainty is modeled by a f...
详细信息
In this work, a mathematical model for stochastic uncertain systems where the system uncertainty is handled by polynomial chaos method is developed. For uncertain systems where the system uncertainty is modeled by a first order polynomial chaos expansion, the estimation of the system states are done by an augmented Kalman filter equations developed by averaged least square method. The performance of the proposed robust estimation algorithm is shown by an uncertain system used as a framework example in previous works.
This study proposes an improved multi-Dimensional Hough Transform technique for the detection of low SNR targets (dim targets) in radar data. The proposed Track-Before-Detect technique improves the multi-Dimensional H...
详细信息
This study proposes an improved multi-Dimensional Hough Transform technique for the detection of low SNR targets (dim targets) in radar data. The proposed Track-Before-Detect technique improves the multi-Dimensional Hough Transform by limiting the target's maximum velocity and incorporating the SNR values of the targets in the algorithm. In addition, the performance is enhanced by confirming the Hough Transform results with a score-based confirmation algorithm.
In this study, a new methodology for combining probability masses from different sources is proposed for Dempster-Shafer theory. Unlike the existing works in the literature, this methodology treats the combination pro...
详细信息
In this study, a new methodology for combining probability masses from different sources is proposed for Dempster-Shafer theory. Unlike the existing works in the literature, this methodology treats the combination problem as an optimization problem and proposes an objective function that uses conflict and entropy measures to solve this problem. The proposed objective function aims to minimize the conflict between the combined masses and the masses comes from the sources, and at the same time maximize the entropy of the combined probability masses. Thus, the difference between the combined probability masses and the masses coming from the sources is minimized while being cautious and avoiding a final certain decision. This new methodology is tested in the mATLAB environment and compared with the existing methods.
In this study, the mutual information between the state sequence and the measurements is proposed as an observability measure and this measure is analyzed in detail for linear time invariant discrete-time Gaussian sys...
详细信息
In this study, the mutual information between the state sequence and the measurements is proposed as an observability measure and this measure is analyzed in detail for linear time invariant discrete-time Gaussian systems. The following results are found as the two basic properties of the measure from the analyses; the unobservable states of the deterministic system have no effect on it. So the measure proposed here is an observability measure of observable states. Secondly, any observable part with no measurement uncertainty makes the measure infinite.
In this work, the concept of sensor management is introduced and stochastic dynamic programming based resource allocation approach is proposed to track multi target. The core of this approach is to use Lagrange relaxa...
详细信息
In this work, the concept of sensor management is introduced and stochastic dynamic programming based resource allocation approach is proposed to track multi target. The core of this approach is to use Lagrange relaxation for decreasing the state space dimension. By this approximation, the overall problem is separated into components instead of using joint markov model to optimize large scale stochastic control problem. The aim of this study is to adaptively allocate radar resources in an optimal way in order to maintain track qualities for multi-target case. Time scale is divided into two levels that are called as micro management and macro management. During this work, we deal with macro management part that aims to construct a policy which is optimal for a given objective function under the resource constraints. In this work, some rule based techniques are added into simulation. The performance of algorithm is analyzed on the average number of update decision and average number of target drops in time horizon.
This work proposes a maximum a posteriori (mAP) based parameter learning algorithm for acoustic-to-articulatory inversion. Inversion method is based on single global linear dynamic system (GLDS) representation of acou...
详细信息
This work proposes a maximum a posteriori (mAP) based parameter learning algorithm for acoustic-to-articulatory inversion. Inversion method is based on single global linear dynamic system (GLDS) representation of acoustic and articulatory data. mAP based learning algorithm considers a prior distribution for the parameter set as well as the likelihood of the training data. Therefore in this paper, we investigate the selection of prior distributions with hyperparameters for GLDS to improve the performance of articulatory inversion. The performance of the proposed learning algorithm and comparison of it with the maximum likelihood (mL) based learning method are examined on an extensive set of examples. These results show that the performance of the articulatory inversion method based on GLDS is significantly improved via mAP based learning algorithm.
暂无评论