Please use this identifier to cite or link to this item: http://rportal.lib.ntnu.edu.tw:80/handle/77345300/32185
Title: 夜視型自主式群組校園巡邏機器人之研究,--子計畫三:探索室外環境功能之導航自走車之研製(II)
Authors: 國立臺灣師範大學電機工程學系
盧明智
許陳鑑
簡忠漢
Issue Date: 31-Jul-2012
Publisher: 行政院國家科學委員會
Abstract: 本計畫為整合型計畫「夜視型自主式群組校園巡邏機器人之研究」之子計畫三。整合型計畫的總體目標為設計一群組機器人,使其能夠自主地在日間(使用一般影像辨識技術)、夜間或是光線不佳(整合紅外線熱像儀影像辨識技術)的環境下,進行環境巡邏的工作。故本子計畫之題目為「探索室外環境功能之導航自走車之研製」,即希望研製一部適合於社區室外環境(住宅區、工廠廠區、學校等)移動之自走輪型機器人(簡稱自走車),使其能夠依據一般影像辨識技術(日間)、或整合紅外線熱像儀影像辨識技術(夜間或是光線不佳時),進行社區環境巡邏的工作。 本研究中著重於發展機器人視覺感測技術為主,本計畫預計自行研發一適合室外環境移動之差動式驅動之自走機器人,搭載多組的攝影機,包括:紅外線攝影機、Pan-Tilt CCD攝影機及搭配其它感測設備如:GPS、電子羅盤、超音波感測器與雷射距離感測器等以蒐集環境的資訊。使用IPM技術,分別定位日、夜間的路面、道路邊線、牆壁與路燈,以提供自走車自我定位。並搭配上GPS與電子羅盤的定位,擷取影像上顯著的特徵,推測導航方式來定位自走車的絕對位置。知道自走車的絕對位置之後, 再應用SIFT 特徵或Harris-Laplace 特徵偵測技術建立特徵資料庫,並依據所建立之影像特徵資料庫,以最小方差法與Kalman Filter 方法進行比對以修正定位誤差,達成自我定位與目標定位,並使用IBDMS(Image-Based Distance Measurement System)的技術來計算物體之間距離並完成三度空間的定位。最後,發展以白天追蹤道路邊線以達成沿路行進及夜間偵測紅外線被動式反射型標線並追蹤標線行進之影像導航控制器。並與各子計畫整合達成日、夜間之巡航任務。
In this project we design an autonomous wheeled robot for day and night patrol of an outdoor campus environment, mainly focusing on integration of cameras and various kinds of sensors to recognize salient landmarks, build the environment map and navigate around the environment autonomously. For this purpose, we will reconstruct a mobile robot equipped with ultrasonic sensors, GPS, digital compass, inertial sensors, laser range finder, IBDMS and IR/CCD camera system to gather environmental information. Then various visual techniques will be developed to recognize landmarks. For example, the color or texture feature can be used to classified cement, grass, or red brick areas, which helps determine the pavement and road edge. Fusion of multiple ultrasonic sensors and cameras may facilitate the detection of walls of a building. With IR images, the lamps can be easily detected by threshold at gray level of image pixels. Based on the rough positioning by GPS and digital compass, the position errors can be further reduced by detecting and tracing salient features, like SIFT features or Harris-Laplace features, with least square error fitting or Kalman filter method, and calculate the real distance between any two objects to achieve a three- dimensional localization by IBDMS method. Then, by deriving inverse sensor model of various sensors, the probabilistic grid map method can be applied to build the environment map. Finally, Then by utilizing the feedback linearization technique, a visual servo controller will be developed to successfully navigate the patrol robot moving along the road edge or artificial IR reflective lane markings.
URI: http://rportal.lib.ntnu.edu.tw/handle/77345300/32185
Other Identifiers: ntnulib_tp_E0607_04_018
Appears in Collections:教師著作

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.