請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/45338
標題: | 自主式機器人之鋪面破損檢測策略 Strategies for Autonomous Robot to Inspect Pavement Distresses |
作者: | Yuan-Hsu Tseng 曾源緒 |
指導教授: | 康仕仲(Shih-Chung Kang) |
關鍵字: | 破損,機器人式,策略,隨機行走,地圖,視覺,虛擬環境, distress,robotic,strategies,random-walk,map,vision,virtual environment, |
出版年 : | 2010 |
學位: | 碩士 |
摘要: | 破損檢測是鋪面維護與修繕作業的重要工作項目。由於鋪面檢測作業需要大量的人力,許多研究者開始發展自動式與機器人式的檢測方法,以提升檢測的效率與精確度。在本研究中,我們致力於發展使用機器人進行檢測之策略。我們發展三種檢測策略。策略一為隨機行走:機器人在受限制的環境內隨機移動並檢測。策略二為隨機行走輔以地圖記錄:機器人在受限制的環境內隨機移動檢測,同時記錄行走過的區域資訊提供機器人進行路徑規劃。策略三賦予機器人視覺能力:機器人透過視覺資訊反應式地調整運動路徑。
為驗證所發展之三種策略,我們以虛擬環境開發測詴域。該測詴域包含五種類型之破損,包含一鱷魚裂縫、一修補、一破洞、一矩形與一圓形人孔。同時,我們亦研發一能自主行駛於測詴域之虛擬機器人。我們利用該機器人執行三種檢測策略並與傳統縱向檢測之成果進行成效比較。成果顯示,使用策略一進行檢測能增加機器人經過破損之頻率,意即使用該策略能比使用傳統縱向檢測偵測�收集到更多的破損資料。使用策略二,顯示利用地圖記錄引導隨機行走能提升成果穩定性;使用策略三,機器人能於時間之內發現更多的破損,能利用視覺能力調整運動路徑以提升發現破損之效率。 Distress inspection is an important task in pavement maintenance. Because pavement inspection requires tremendous human resources, many investigators start developing automatic and robotic inspection methods to increase the efficiency and accuracy. In this research, we specific focus on developing strategies for executing the inspection tasks using robots. We developed three strategies. The first strategy is random-walk. Robot surveys randomly in a confined environment. The second strategy is random-walk with map recording. Robot randomly wanders with recording the information it has gone through. The third one adds the vision capacity to the robot. Robot determines inspection path reactively based on the visual information. To validate the three strategies, we developed a test field in a virtual environment. This test field includes 5 type of distress, including an alligator crack, a patching, a breaking hole, a rectangular manhole and a circular manhole. We also developed a virtual robot which can autonomously navigate in the test field. We then implemented the three survey strategies in the robot and compare their performances with traditional longitudinal survey method. The results show that using the first strategy, we can increase frequency for passing the distresses; it means that robot can detect and collect data more times than traditional longitudinal survey. The results of the second strategy show that we can increase the repeatability by using map recording to guide random-walk. The results of third strategy show that the robot can find more distresses in a certain amount of time; it means that we can improve survey efficiency by adding vision capacity to adjust motion path when distresses detected. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/45338 |
全文授權: | 有償授權 |
顯示於系所單位: | 土木工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-99-1.pdf 目前未授權公開取用 | 1.9 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。