| Summary: | 博士 === 國立高雄第一科技大學 === 工程科技研究所 === 100 === Abstract
In modern tracking analysis, many researchers have attempted the various models of the target tracking process. General, target tracking is the important problem in military and civilian fields. Many mathematical models have been formulated to manipulate the target tracking system in the world. In the mid 1950’s, relatively simple α-β and α-β-γ filter trackers were developed to deal with this problem. Their advantages are simple computation and quick response when target maneuvering. Nevertheless, it is to be more adaptable under complicated situation in multi-target tracking or changeable target. The achievements of the dissertation are as follows:
1. To further predict the acceleration and improve the tracking accuracy, an additional state value called jerk that is time derivative of acceleration will need to be observed. As a result, it exhibits significant improvement in tracking accuracy over the α-β-γ filter.
2. Not unexpectedly, however, the new α-β-γ-δ filter takes more computation time in the optimization process. To overcome this weakness, an optimal simulation technique via Genetic Algorithm (GA) is proposed. The developed GA-based α-β-γ-δ filter finds not only the optimal set of filter parameters to minimize position tracking errors but could also reduce the computation time in some time steps.
3. Besides, two heuristic methods (one is GA method, the other is Taguchi method.) are combined together for improving tracking accuracy. Finally, the result shows it leads to get more tracking accuracy.
4. Furthermore, the Evolutionary Programming (EP) algorithm is then introduced. As a result, the developed EP-based α-β-γ-δ filter also finds not only the optimal set of filter parameters to minimize position tracking errors but could also reduce the computation time.
5. In this paper, it is also to compare the tracking accuracy between GA-based α-β-γ-δ filter and EP-based α-β-γ-δ filter. The result shows the same tracking accuracy.
6. The applications of others shall be explored in future.
As above mention, another algorithm method can be explored in target tracking process for optimization in future.
Keywords: target tracker; Genetic Algorithm; the α-β-γ-δ filter; the α-β-γ filter; the GA-based α-β-γ-δ filter; Evolutionary Programming; the EP-based α-β-γ-δ filter
|
|---|
