臺灣博碩士論文加值系統

無人飛行載具（Unmanned Aerial Vehicle, UAV）近年來蓬勃發展，而四軸飛行器是一種直升機式的無人飛行載具，本論文以四軸飛行器為例，進行四軸飛行器的動態模型之模擬、觀測與探討，雖然四軸飛行器的動態模型運動方式非常複雜，而觀測方式卻很簡易，只需要一台攝影機或是一位觀察員直接觀測其位置。量測結果以角度呈現，這種以角度去描述物體位置的追蹤方法可視為“僅方向追蹤”的問題(bearings-only tracking)，而具體描述方法，又以柱座標最為適合。一般來說，四軸飛行器操作模式為升空後，在同一個水平面上移動，利用柱座標的特性，將複雜的三維空間物體追蹤，視為一個二維的平面物體追蹤，本論文結合粒子濾波器演算法，在平面上追蹤物體。
追蹤的演算法不僅僅只有粒子濾波器，還有卡爾曼濾波器、擴增卡爾曼濾波器…等等。但，四軸飛行器的動態模型並非線性方程式，甚至狀態轉移矩陣裡面有未知數，無法將矩陣做反矩陣，故本論文選擇粒子濾波器。
關鍵字：粒子濾波器, 物件追蹤

The Unmanned Aerial Vehicle (UAV) is a popular topic in recent years, and the quadcopter is a helicopter-kind unmanned aerial vehicle. This paper uses quadcopter as an example to simulate and observe the dynamic model of quadcopter. Although the dynamic model of the quadcopter is very complicated, but the observation method is very simple, only using a camera or an observer to directly observe its position. The measurement results represent as an angle. This tracking method can be deal as a “bearings-only tracking” problem, we use the cylindrical coordinate which is a proper method. Generally, the quadcopter is to move horizontally in the air. It simplify the complex three-dimensional object tracking as a two-dimensional object tracking.
here are many tracking algorithm, like Kalman filter, Extend Kalman filter and adaptive filter. However, the dynamic model of the quadcopter is not a linear equation, and the state transition matrix, have unknowns parameters, and the matrix cannot be inverse . Therefore, this paper combine the particle filter to tracking object on plane.
Keywords: particle filter, object tracking

Table of Contents
中文摘要 iii
Abstract iv
Acknowledgement v
Table of Contents vi
List of Tables vii
List of Figures viii
Chapter 1 Introduction ix
Chapter 2 Quadcopter Position Tracking 1
2.1 Quadcopter’s dynamic model 1
2.2 Measurement with the angle 2
2.3 Bearings-only problem 3
Chapter 3 Object Position Tracking on Plane with the Particle Filter 4
3.1 Position tracking base on the bayes filter 4
3.2 A method for the position tracking on the plane with the particle filter 8
3.3Measurement improvement with the Angular velocity 10
Chapter 4 Simulation Results and Discussion 11
4.1 Simulation setup 11
4.2 Simulation results and discussion 13
Chapter 5 Conclusion 15
Reference 16

Table of Contents
中文摘要iii
Abstractiv
Acknowledgementv
Table of Contentsvi
List of Tablesvii
List of Figuresviii
Chapter 1 Introductionix
Chapter 2 Quadcopter Position Tracking1
2.1 Quadcopter’s dynamic model1
2.2 Measurement with the angle2
2.3 Bearings-only problem3
Chapter 3 Object Position Tracking on Plane with the Particle Filter4
3.1 Position tracking base on the bayes filter4
3.2 A method for the position tracking on the plane with the particle filter8
3.3Measurement improvement with the Angular velocity10
Chapter 4 Simulation Results and Discussion11
4.1 Simulation setup11
4.2 Simulation results and discussion13
Chapter 5 Conclusion15
Reference16