Tzvetan Semerdjiev

A study of a target tracking algorithm using global nearest neighbor approach

This paper compares two algorithms for Multiple Target Tracking (MTT), using Global Nearest Neighbor (GNN) and Suboptimal Nearest Neighbor (SNN) approach respectively. For both algorithms the observations are divided in clusters to reduce computational efforts. For each cluster the assignment problem is solved by using Munkres algorithm or according SNN rules. Results reveal that in some cases the GNN approach gives batter solution than.SNN approach. The computational time, needed for assignment problem solution using Munkres algorithm is studied and results prove that it is suitable for real time implementations.

Application of a Monte Carlo method for tracking maneuvering target in clutter

The Monte Carlo methods provide a possibility for improved sub-optimal Bayesian estimation. In preceding studies the authors have suggested a new implementation of the general bootstrap simulation approach — the bootstrap multiple model (BMM) filter for tracking a maneuvering target. In the present paper this algorithm is further extended for operating in a cluttered environment. Probabilistic data association (PDA), taking into account the possible measurement-to-target association hypotheses, is incorporated into the BMM algorithm to overcome the measurement–origin uncertainty. By simulation the proposed BMM PDA algorithm is evaluated and compared with the well-known interacting multiple model (IMM) PDA filter. The obtained results demonstrate a superior tracking performance of the BMM PDA algorithm at the cost of an increase in computation.

Monte Carlo algorithm for ballistic object tracking with uncertain drag parameter

The problem of tracking a reentry ballistic object by processing radar measurements is considered in the paper. Sequential Monte Carlo-based filter is proposed for dealing with high nonlinearity of the object dynamics. A multiple model configuration is incorporated into the algorithm for overcoming the uncertainty about the object ballistic characteristics. The performance of the suggested multiple model particle filter (PF) is evaluated by Monte Carlo simulation.

A study of clustering applied to multiple target tracking algorithm

In this paper the effectiveness of two Data Association algorithms for Multiple Target Tracking (MTT) based on Global Nearest Neighbor approach are compared. As the time for assignment problem solution increases nonlinearly depending on the problem size, it is useful to divide the whole scenario on small groups of targets called clusters. For each cluster the assignment problem is solved by using-Munkres algorithm. Results reveal that the computational time especially for large scenarios decreases significantly when clustering is used.

Estimation of target behavior tendencies using dezert-smarandache theory

This paper presents an approach for tar- get behavior tendency estimation (Receding, Approach- ing). It is developed on the principles of Dezert- Smarandache theory (DSmT) of plausible and para- doxical reasoning applied to conventional sonar ampli- tude measurements, which serve as an evidence for COT- responding decision-making procedures. In some real world situations it is dificult to finalize these proce- dures, because of discrepancies in measurements inter- pretation. In these cases the decision-making process leads to conflicts, which cannot be resolved using the well-know methods. The aim of the perjomed study is to present and to approve the ability of DSmT to fi- nalize successfully the decision-making process and to assure awareness about the tendencies of target behav- ior in case of discrepancies in measurements interpre- tation. An example is provided to illustrate the bene- fit of the proposed approach application in comparison of fuzzy logic approach, and its ability to improve the overall tracking perfonnance.

Monte Carlo Algorithm for Maneuvering Target Tracking and Classification

This paper considers the problem of joint maneuvering target tracking and classification. Based on the recently proposed particle filtering approach, a multiple model particle filter is designed for two-class identification of air targets: commercial and military aircraft. The classification task is implemented by processing radar (kinematic) measurements only, no class (feature) measurements are used. A speed likelihood function for each class is defined using a priori information about speed constraints. Class-dependent speed likelihoods are calculated through the state estimates of each class-dependent tracker. They are combined with the kinematic measurement likelihoods in order to improve the process of classification. The performance of the suggested multiple model particle filter is evaluated by Monte Carlo simulations.

Target tracking using Monte Carlo simulation

This paper proposes and studies an implementation of the bootstrap stochastic simulation approach for estimating a hybrid system – the bootstrap multiple model (BMM) algorithm. The BMM filter is applied to tracking maneuvering target. The tracking capabilities of the filter are demonstrated by computer simulation.

About the Specifics of the IMM Algorithm Design

It is well known, that interacting multiple model (IMM) state estimation algorithm is one of the most cost-effective filters for tracking maneuvering targets. The present paper is related to the specifics of the IMM algorithm design. It combines the results, conclusions and experience of different authors considered in their papers. The results discussed and depicted here are root mean square errors and the filters ability to distinct the various flight phases. This paper helps the air traffic control experts fast and easy to make a decision which IMM configuration is suitable for a given problem.

On-Line State Estimation of Maneuvering Objects by Sequential Monte Carlo Algorithm

A stochastic sampling algorithm for recursive state estimation of nonlinear dynamic systems is designed and realized in this study. It is applied to the problem of tracking two maneuvering air targets in the presence of false alarms. The performance of the proposed algorithm is evaluated via Monte Carlo simulation. The results show that the nonlinear Bayesian filtering can be efficiently accomplished in real time by simple Monte Carlo techniques.