跳到主要內容

臺灣博碩士論文加值系統

(3.236.68.118) 您好!臺灣時間:2021/07/31 19:08
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:王伶文
研究生(外文):Wang, Ling-Wen
論文名稱:自發性隊伍行進系統
論文名稱(外文):The study on an Autonomous Formation Marching System
指導教授:黃國勝黃國勝引用關係
指導教授(外文):Kao-Shing Hwang
口試委員:蔡清池李祖聖林金玲黃國勝
口試委員(外文):Ching-Chih TsaiTzuu-Hseng S. LiJin-Ling LinKao-Shing Hwang
口試日期:2011-07-27
學位類別:碩士
校院名稱:國立中正大學
系所名稱:光機電整合工程研究所
學門:工程學門
學類:機械工程學類
論文種類:學術論文
論文出版年:2011
畢業學年度:99
語文別:英文
論文頁數:56
中文關鍵詞:融合學習行為隊伍行進自發性
外文關鍵詞:FBQL(Fusion Approach Based on Q-Learning)Formation marchingAutonomous
相關次數:
  • 被引用被引用:0
  • 點閱點閱:186
  • 評分評分:
  • 下載下載:29
  • 收藏至我的研究室書目清單書目收藏:0
本論文主要研究重點在於增加多台機器人在隊形行進時的自主的能力,依據學姊先前所設計的環境架構和行為學習演算法為基礎,利用Webots去模擬隊伍在行進中可能發生的突發情況,增加每隻機器人在遇到此情況的應對反應,使得機器人的自主性能夠更為提升,進而改善在實作時可能發生影響隊伍行進的情況。將以兩種隊形行進中可能發生的情況來做改善和增加其自助能力,第一、假設在隊伍行進中,有一隻機器人發生故障,在一開始設定發生故障的機器人及發生故障的時間,系統開始運作後,當故障情況發生時其他機器人能夠即時反應重新安排位置繼續完成任務。第二、當某隻機器人因電力不足等情況發生脫隊現象,其他機器人能夠維持隊形而等候脫隊的機器人歸隊再繼續前進,或是捨棄脫隊機器人,以維持隊形完成到達終點的任務為優先。
在本架構中使用Fusion Approach Based on Q-Learning (行為融合學習)來做為每台機器人應具備的行為學習能力,而針對這個目的,分別設計了跟隨距離行為,跟隨角度行為,避障行為以便達到避開障礙物並且維持完整隊形抵達目標點的這個目的,在每台機器人僅知自己以及他台機器人的座標位置的情況下,只需在一開始給予其目的地座標以及何種隊形,即可自主的判斷出各自在隊形中的位置。即使不給定個體機器人過多複雜的已知條件,仍可以利用簡單自主的能力達成多台機器人智慧型地完成隊形且到達目地點的表現。
The main objective of the thesis is to study on adding the multiple robots’ individual abilities in formation marching system. The environment structure and behavior of learning algorithms are based on the design of previous works. Using WEBOTS simulates the condition of emergencies happened in formation marching. Adding the each robot’s correspond reaction by programming to increase the individual abilities of robots in formation. Using simulation result improves the impact factors in formation marching of implementation.
Using two conditions of formation marching may occur, and then the system will try to improve it and increase the individual abilities. First, one of robots is crash down in formation marching. To set the broken-down robot and the broken-down time in the beginning so that the setting robot will be crash down in formation marching. When the accident happened, the system in order to continue to complete the task then the robots of system will assign positions once more. The other robots can immediate response when accident happened. Second, when one of robot is low power or other factor happened lead to robot is off the team. The system will determine the distance between the off team robot and leader. The system will determine whether to keep formation and to wait for the robot returns or to give up the off team robot and to finish the task that keeps going forward to goal first.
The algorithm used in the thesis is Fusion Approach Based on Q-Learning which has been evolved for several generations in the lab. We designed three behaviors including the behaviors of following distance, following angle, and obstacle avoidance so that the robots can reach the target with complete formation safely.
The robot can judge the situation independently and safely where it should go in the formation under the condition that the robot only knows owns and other teammates’ coordinates as long as the destination coordinates and types of formations are given to the team.
中文摘要 i
ABSTRACT ii
TABLE OF CONTENTS iv
LIST OF FIGURES vi
I. INTRODUCTION 2
1.1 Multi-robot 2
1.2 Motivation and Purpose 3
1.3 Chapter Organization 4
II. BACKGROUND 5
2.1 Reinforcement Learning 5
2.2 Q-Learning 7
2.3 Fusion Approach Based on Q-Learning(FBQL) 8
2.4 Hungarian Alorithm 10
2.5 Assignment of ID 12
2.6 Shape Representation 14
III. SIMULATION and SYSTEM ARCHITECTURE 16
3.1 Similation Introduction 16
3.2 Behaviors Design 19
3.2.1 The Behavior of Following Distance 19
3.2.2 The Behavior of Following Angle 21
3.2.3 The Behavior of Obstacle Avoidance 23
3.2.4 The Behavior of Break down 25
3.2.5 The Behavior of off the Team 27
3.3 Architecture of System 29
IV. IMPLEMENTATION and RESULTS 34
4.1 Description of Simulation Conditions 34
4.2 Condition 1:One of Robots is Crash down 34
4.3 Condition 2:One of Robots is Off The Team 45
V. CONCLUSIONS and FUTURE WORK 46
REFERENCE 47
[1]Jakob Fredslund and Maja J. Mataric, “A General Algorithm for Robot Formations Using Local Sensing and Minimal Communication,” IEEE Transactions and Automation, VOL. 18, NO. 5, 2000.
[2]Tucker Balch and Ronald C. Arkin, “Behavior-Based Formation Control for Multirobot Teams,” IEEE Transaction on Robotics and Automation, VOL. 14, NO. 6, 1998.
[3]Petter Bgren and Naomi Ehrich Leonard, “Obstacle Avoidance in Formation,” IEEE International Conference on Robotics & Automation, 2003.
[4]Jaydev P. Desai, James P. Ostrowski, and Vijay Kumar, “Modeling and Control of Formations of Nonholonomic Mobile Robots,” IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, VOL. 17, NO. 6, 2001.
[5]Petter Bgren and Naomi Ehrich Leonard, “,Obstacle Avoidance in Formation,” IEEE International Conference on Robotics & Automation, 2003.
[6]Herbert G. Tanner and Amit Kumarr, “Towards Decentralization of Multi-robot Navigation Functions,” IEEE International Conference on Robotics and Automation, 2005.
[7]Alan Wagner and Ronald Arkin, “Towards Multi-Robot Communication-Senisitive Reconnaissance,” IEEE International Conference on Robotics & Automation,2004
[8]C. J. Wu, “A Behavior Fusion Approach Based on Q-Learning for Mobile robots,” Master thesis, National Chung Cheng University, Taiwan, 2006.
[9]Ya-Ling Wang, “Adaptive Formation of Multiple Robots Based on Behavioral Learning,” Master thesis, National Chung Cheng University, Taiwan, 2008.[Text in Chinese]
[10]R. S. Sutton and A. G. Barto, “Reinforcement Learning,” The MIT Press, Cambridge, Massachusetts, London, England, 1998.

QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
無相關論文