請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/90583
標題: | 具最小變形引導之分散式模型預測控制之無人機編隊控制系統 Distributed Model Predictive Control with Minimum Deformation Guidance for Multi-UAV Formation Control System |
作者: | 陳有文 Yu-Wen Chen |
指導教授: | 傅立成 Li-Chen Fu |
關鍵字: | 路徑規劃,編隊控制,分散式模型預測控制,多無人機系統,障礙空間, motion planning,formation control,distributed model predictive control,multi-UAV system,obstacle environments, |
出版年 : | 2023 |
學位: | 碩士 |
摘要: | 本論文探討在障礙環境中的多無人機(multi-UAV)的編隊控制任務。編隊控制任務包含三個主要目標:到達目的地、維持編隊形態和避免碰撞。近年來有許多編隊控制系統使用優化框架來整合不同的控制目標,但基於梯度的優化方法通常只能獲得局部最優解,並且缺乏全局規劃能力。為了應對這一挑戰,我們提出了一種專為編隊控制而設計的路徑規劃演算法,並將其集成到多無人機編隊控制系統中。該系統由兩個主要組成部分組成。首先,我們開發了一種編隊路徑規劃算法,能夠為編隊中心生成安全的參考路徑,並為每個無人機的尋找安全航點並保有最小化的編隊形態的變形。其次,我們引入了分布式模型預測控制器(MPC)用於具有非線性動態特性的四旋翼飛行器。這種模型預測控制器能夠使無人機追蹤由路徑規劃得來的參考軌跡,並通過鄰近無人機提供的訊息維持編隊形態。此外,與許多假設已知環境並僅通過模擬驗證的現有研究不同,我們設計了一種能夠處理感測器數據、對周圍障礙物進行建模的演算法,並且在現實環境中實現提出的編隊控制系統。我們在論文的最後用多樣的模擬場景與實際實驗驗證了我們提出的編隊控制系統的效果與可行性。 The thesis considers the formation control task for multiple unmanned aerial vehicles (multi-UAV) in obstacle environments. This task encompasses three primary objectives, which are target reaching, formation maintenance, and collision avoidance, respectively. While most recent formation control systems employ optimization frameworks to synthesize control objectives, gradient-based optimization methods often yield only locally optimal solutions and lack global planning capabilities. To tackle this challenge, we propose a planning algorithm specifically designed for formation control, integrating it into the proposed multi-UAV formation control system. The system comprises two main components. Firstly, we develop a formation path planning algorithm capable of generating a safe reference path for the formation center and determining secure waypoints for each agent to minimize formation pattern deformation. Secondly, we introduce a distributed model predictive controller (MPC) for the quadrotor agents, which possess nonlinear dynamics. This MPC controller facilitates agent tracking of the reference trajectory calculated by the planner and maintains the formation based on information from neighboring agents. Furthermore, unlike many previous works that assume a known environment and only validate through simulation, we design our algorithm to process sensor data, model surrounding obstacles, and implement the proposed formation control system in real-world. Finally, we conduct simulation examples and real-world experiments in diverse scenarios to showcase the feasibility and effectiveness of the proposed formation control system. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/90583 |
DOI: | 10.6342/NTU202302307 |
全文授權: | 同意授權(全球公開) |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-111-2.pdf 此日期後於網路公開 2026-08-04 | 15.49 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。