一種支撐式管道檢測(cè)機(jī)器人的結(jié)構(gòu)和性能優(yōu)化設(shè)計(jì)含6張CAD圖

一種支撐式管道檢測(cè)機(jī)器人的結(jié)構(gòu)和性能優(yōu)化設(shè)計(jì)含6張CAD圖,一種,支撐,管道,檢測(cè),機(jī)器人,結(jié)構(gòu),性能,優(yōu)化,設(shè)計(jì),CAD 雙輪鏈平面管道檢測(cè)機(jī)器人 摘要 介紹了一種新型的多傳感器管道檢測(cè)機(jī)器人，用于80-100mm管道的檢測(cè)。該機(jī)器人的特點(diǎn)是只需使用兩個(gè)輪鏈即可實(shí)現(xiàn)驅(qū)動(dòng)和轉(zhuǎn)向功能。與普遍采用的三輪鏈條管道機(jī)器人相比，新設(shè)計(jì)允許簡(jiǎn)單的機(jī)器人控制和方便的用戶界面，特別是在T形分支。作為另一個(gè)優(yōu)點(diǎn)，這種機(jī)器人的平面形狀允許在機(jī)器人的兩側(cè)安裝額外的傳感器。介紹了系統(tǒng)的運(yùn)動(dòng)學(xué)和三種控制方式。最后，通過(guò)實(shí)驗(yàn)驗(yàn)證了該機(jī)器人系統(tǒng)的性能。 關(guān)鍵詞:管道機(jī)器人；系統(tǒng)；運(yùn)動(dòng)學(xué) I.緒論 管道檢測(cè)機(jī)器人的功能可以描述為驅(qū)動(dòng)、轉(zhuǎn)向、檢測(cè)和檢索。 而用于直徑小于100mm管道檢測(cè)的機(jī)器人，在設(shè)計(jì)緊湊的轉(zhuǎn)向機(jī)構(gòu)和安裝磁探頭、超聲波探頭等傳感器檢測(cè)裂紋、破裂、泄漏等方面存在特殊困難。 管道機(jī)器人機(jī)構(gòu)在機(jī)器人技術(shù)領(lǐng)域有著悠久的發(fā)展歷史，按其運(yùn)動(dòng)模式可分為幾種基本形式。它們有輪式、尺蠖式、腿式、螺旋式、履帶式和被動(dòng)式。其中輪式管道檢測(cè)機(jī)器人最為流行，[1]-[8]。然而，它們不適合在垂直路徑或在t分支操作。近10年來(lái)，人們對(duì)差動(dòng)驅(qū)動(dòng)型機(jī)構(gòu)[9]-[11]進(jìn)行了較為深入的研究。差動(dòng)驅(qū)動(dòng)類型通常有三個(gè)動(dòng)力輪鏈。通過(guò)獨(dú)立控制每個(gè)鏈條的速度，機(jī)器人可以通過(guò)肘部和t型分支。然而，當(dāng)只使用一個(gè)機(jī)器人模塊時(shí)，有時(shí)會(huì)在T支[9]處發(fā)生奇異運(yùn)動(dòng)。為了解決這一問(wèn)題，已經(jīng)開(kāi)發(fā)了幾種方法，如主動(dòng)轉(zhuǎn)向關(guān)節(jié)機(jī)構(gòu)[12-13]或兩個(gè)機(jī)器人模塊[9]協(xié)作。然而，整個(gè)機(jī)器人系統(tǒng)的體積變得龐大。使用三個(gè)動(dòng)力輪鏈的另一個(gè)缺點(diǎn)是沒(méi)有足夠的空間在機(jī)器人體內(nèi)安裝更多的傳感器，因?yàn)槿齻€(gè)輪鏈占據(jù)了管道的大部分橫截面積，特別是直徑小于100mm的管道。目前，機(jī)器人身體前只安裝了一個(gè)攝像頭。T. Okad等[14-16]開(kāi)發(fā)了平板式管道檢測(cè)機(jī)器人。然而，他們的設(shè)計(jì)是復(fù)雜的，并用于大型管道。 針對(duì)這些因素，我們提出了一種雙動(dòng)力輪鏈的管道檢測(cè)機(jī)構(gòu)。兩個(gè)輪鏈以180度的角度分開(kāi)布置，所以可以在機(jī)器人身體的兩側(cè)附加傳感器。各輪采用兩臺(tái)電機(jī)控制;一種用于駕駛，另一種用于駕駛。一組計(jì)數(shù)車輪創(chuàng)建一個(gè)螺旋運(yùn)動(dòng)，以實(shí)現(xiàn)轉(zhuǎn)向，另一組車輪創(chuàng)建一個(gè)直線運(yùn)動(dòng)在管道內(nèi)，沒(méi)有任何奇異的運(yùn)動(dòng)彎頭或t型分支。 該機(jī)器人系統(tǒng)由3種控制模式組成;移動(dòng)、探測(cè)和搜索。我們將為每種模式引入相應(yīng)的策略。機(jī)器人機(jī)構(gòu)采用并聯(lián)機(jī)構(gòu)設(shè)計(jì)，使其具有可折疊特性?？烧郫B的特性允許適應(yīng)的車輪機(jī)制，以墻壁內(nèi)的管道。第二部分介紹了機(jī)器人系統(tǒng)的特點(diǎn)。運(yùn)動(dòng)學(xué)分析在第三節(jié)中介紹。第四部分通過(guò)仿真和實(shí)驗(yàn)驗(yàn)證了該機(jī)器人系統(tǒng)的有效性，最后給出了結(jié)論. II.機(jī)器人的特點(diǎn) A.整個(gè)系統(tǒng)和機(jī)器人設(shè)備 圖一機(jī)器人系統(tǒng)由控制箱和機(jī)器人裝置組成。該機(jī)器人系統(tǒng)由一個(gè)操縱桿界面控制。機(jī)器人用戶可以利用視覺(jué)信息對(duì)管道的狀態(tài)進(jìn)行檢測(cè)。 該機(jī)器人裝置由一個(gè)主體和兩個(gè)輪鏈組成，如圖2所示，機(jī)器人長(zhǎng)度80mm，外徑100mm。 圖1帶兩輪鏈機(jī)制的管道檢測(cè)機(jī)器人系統(tǒng) 該機(jī)器人機(jī)構(gòu)可以在三種不同的模式下工作;驅(qū)動(dòng)模式、檢測(cè)模式、搜索模式。在驅(qū)動(dòng)模式下，機(jī)器人處于運(yùn)動(dòng)狀態(tài)。在檢測(cè)模式下，機(jī)器人檢測(cè)管道內(nèi)壁的狀態(tài)。在搜索模式下，機(jī)器人收 圖2管道檢測(cè)機(jī)器人系統(tǒng)的整個(gè)結(jié)構(gòu)具有兩個(gè)輪鏈機(jī)制：（a） 3D 模型， （b開(kāi)發(fā)的機(jī)器人。 B.主體和側(cè)傳感器安裝器 圖三顯示主體由按鍵滑塊、兩個(gè)主體軸、四個(gè)壓縮彈簧、傳感器貼片、前攝像頭和兩個(gè)側(cè)傳感器組成。關(guān)鍵滑塊(移動(dòng)關(guān)節(jié))在主體軸上滑動(dòng)，并連接到輪鏈的連桿上。壓縮彈簧能適應(yīng)鏈輪外徑的變化。這種設(shè)計(jì)允許機(jī)器人身體的可折疊特性。每個(gè)按鍵滑塊由按鍵連接器連接，這樣兩個(gè)輪子就能產(chǎn)生相同的折疊運(yùn)動(dòng)。因此，如果我們使用這種設(shè)計(jì)，主體總是位于管道的中心。 在該機(jī)構(gòu)中，機(jī)器人采用兩鏈的平面形狀。因此，在機(jī)器人身體的左右兩側(cè)都有一些空間，可以安裝各種傳感器，如MT, UT，或視覺(jué)傳感器(攝像機(jī))。在本文中，我們使用前置攝像頭、右攝像頭和圖 3 中顯示的左攝像頭。從每個(gè)攝像機(jī)中，我們可以獲取圖 4 中顯示的管道內(nèi)部信 圖3主體結(jié)構(gòu) 圖4相機(jī)視圖 - 前置攝像頭、左攝像頭和右攝像頭 C.車輪鏈機(jī)制 所述主體與所述輪鏈的折疊機(jī)構(gòu)相連。輪鏈由折疊機(jī)構(gòu)(并聯(lián)機(jī)構(gòu))、驅(qū)動(dòng)電機(jī)(10)、主動(dòng)輪、轉(zhuǎn)向輪(怠速輪)、轉(zhuǎn)向電機(jī)(8)和轉(zhuǎn)向機(jī)構(gòu)組成，如圖2和圖5所示。車輪驅(qū)動(dòng)電機(jī)通過(guò)斜齒輪動(dòng)力傳動(dòng)帶動(dòng)主動(dòng)輪。如圖5的前視圖所示，轉(zhuǎn)向電機(jī)改變方向盤的方向，從而將機(jī)器人轉(zhuǎn)向想要的方向。 III.機(jī)器人運(yùn)動(dòng)規(guī)劃與檢測(cè)功能 A. 管道的基本運(yùn)動(dòng) 圖6直管線的運(yùn)動(dòng)（a）、 前視圖（b） 等軸測(cè)量視圖 （c）頂部視圖 管道檢測(cè)機(jī)器人系統(tǒng)需要通過(guò)直線和曲線管道。要做到這一點(diǎn)，我們需要開(kāi)發(fā)新的運(yùn)動(dòng)規(guī)劃算法。圖6(a)中的xyz表示連接到管道檢測(cè)機(jī)器人本體的局部坐標(biāo)系。我們定義oi為方向盤的轉(zhuǎn)向角度。在沒(méi)有轉(zhuǎn)向角度的情況下，機(jī)器人在管道內(nèi)不作任何旋轉(zhuǎn)，沿z軸前后移動(dòng)。當(dāng)機(jī)器人操作員使用前置攝像頭檢查管道狀態(tài)時(shí)，機(jī)器人的前后移動(dòng)用于驅(qū)動(dòng)模式。 圖7顯示機(jī)器人的旋轉(zhuǎn)運(yùn)動(dòng)。如果兩個(gè)方向盤被安排在相反的方向，如圖7（a），機(jī)器人明智地旋轉(zhuǎn)時(shí)鐘，并創(chuàng)建一個(gè)螺絲釘。 圖7為機(jī)器人的旋轉(zhuǎn)運(yùn)動(dòng)。如圖7(a)所示，如果車輪排列在相反的方向，機(jī)器人會(huì)明智地旋轉(zhuǎn)計(jì)數(shù)器時(shí)鐘。這種螺絲運(yùn)動(dòng)可用于檢測(cè)模式：一旦前置攝像頭檢測(cè)到管道內(nèi)表面的某些缺陷，機(jī)器人就會(huì)使用螺絲運(yùn)動(dòng)定位側(cè)攝像頭的位置，以便側(cè)攝像頭能夠清楚地檢查現(xiàn)場(chǎng)。 通過(guò)控制方向盤在同一方向上的角度，也可以實(shí)現(xiàn)肘部或T-分支的轉(zhuǎn)向運(yùn)動(dòng)。圖 8（a） 顯示，當(dāng)兩個(gè)方向盤轉(zhuǎn)向左方向時(shí)，機(jī)器人在 T 分支處向左轉(zhuǎn)。圖 8 （b） 用于轉(zhuǎn)向正確的方向。需要注意的是，即使在轉(zhuǎn)向運(yùn)動(dòng)中，兩個(gè)驅(qū)動(dòng)電機(jī)的速度也是一樣的。因此，與三個(gè)電動(dòng)車輪機(jī)器人相比，轉(zhuǎn)向的控制力要小一些，三個(gè)車輪具有 差速，可通過(guò)肘部或 T 分支。 (a) (b) 圖8T 分支和肘部的運(yùn)動(dòng) （a） 左轉(zhuǎn)運(yùn)動(dòng) （b） 右轉(zhuǎn)運(yùn)動(dòng)。 B.控制模式 被調(diào)查管道檢測(cè)機(jī)器人具有控制模式;驅(qū)動(dòng)模式、檢測(cè)模式、搜索模式。首先，在驅(qū)動(dòng)模式下，機(jī)器人通過(guò)前置攝像頭前后移動(dòng)來(lái)檢測(cè)管道的狀態(tài)。在檢測(cè)模式下，利用正面和側(cè)面的攝像頭對(duì)墻上的特殊點(diǎn)進(jìn)行精確定位。在搜索模式下，通過(guò)使用側(cè)攝像頭來(lái)識(shí)別光點(diǎn)的狀態(tài)或問(wèn)題。圖9為三種模式的圖片。 圖9實(shí)驗(yàn)結(jié)果： 管道內(nèi)導(dǎo)航 IV、機(jī)制分析 車輪機(jī)構(gòu)的詳細(xì)結(jié)構(gòu)、坐標(biāo)系、關(guān)節(jié)變量及參數(shù)如圖10所示。xyz表示連接到管道的本地坐標(biāo)框架。 圖10折疊機(jī)制 V、實(shí)施 A. 控制器 機(jī)器人控制器由控制盒、控制PC、抓取板、操縱桿界面和機(jī)器人設(shè)備組成，如圖12所示。機(jī)器人的控制是通過(guò)串行通信來(lái)執(zhí)行的。本系統(tǒng)采用單片機(jī)(Atmega128)。單片機(jī)通過(guò)產(chǎn)生PWM信號(hào)來(lái)計(jì)算電機(jī)轉(zhuǎn)速。它可以控制所有的微型直流電機(jī)。所有的電機(jī)驅(qū)動(dòng)器和MCU集成在控制箱內(nèi)。 管道的視圖通過(guò)安裝在機(jī)器人身體前面的微型CMOS攝像頭提供給用戶。這個(gè)攝像模塊可以檢查管道內(nèi)部的情況。裝有攝像頭的機(jī)器人裝置如圖12所示。 B. 機(jī)器人設(shè)備 電機(jī)嵌入在車輪機(jī)構(gòu)的電機(jī)箱中。轉(zhuǎn)向馬達(dá)的峰值扭矩為 30mNm。選擇 Maxon re 8 和 GP 8B 齒輪頭用于轉(zhuǎn)向電機(jī)。驅(qū)動(dòng)電機(jī)的峰值扭矩為150 毫安米。選擇 Maxon re 10 和 GP 10A 齒輪頭用于駕駛電機(jī)。 表一 顯示了機(jī)器人的規(guī)格。機(jī)器人模塊的長(zhǎng)度為80mm，機(jī)器人本體的外徑從80mm到100mm不等。包括攝像機(jī)和照明裝置在內(nèi)的機(jī)器人裝置總長(zhǎng)度為94mm。機(jī)器人的重量是237克。本文采用直徑為100 mm的管道作為試驗(yàn)平臺(tái)。 C. 攝像機(jī)傳感器 本文采用了三種CMOS攝像機(jī)傳感器。其直徑為10mm，最小焦距為2cm，可用于直徑小于100mm的管道中。 VI.實(shí)驗(yàn)結(jié)果 實(shí)驗(yàn)中采用了一種由直管、t型管和彎頭組成的管道。它的內(nèi)徑是100毫米。t形支管和彎頭均為商用產(chǎn)品。 圖13試驗(yàn)臺(tái) 圖13為實(shí)驗(yàn)的試驗(yàn)臺(tái)，圖14為彎頭處的驅(qū)動(dòng)和轉(zhuǎn)向運(yùn)動(dòng)。從圖15、圖16可以看出，管道檢測(cè)機(jī)器人能夠在t支路上進(jìn)行轉(zhuǎn)向，并通過(guò)向后運(yùn)動(dòng)和轉(zhuǎn)向運(yùn)動(dòng)相結(jié)合成功改變方向。通過(guò)實(shí)驗(yàn)驗(yàn)證了該機(jī)器人系統(tǒng)的性能。附件中的視頻片段顯示了實(shí)驗(yàn)結(jié)果。 圖14肘部的運(yùn)動(dòng)規(guī)劃：（a）=（b） 直管線向前運(yùn)動(dòng)，（c） 車輪轉(zhuǎn)向，（d）~（f）肘部向左方向轉(zhuǎn)動(dòng) VII.結(jié)論 針對(duì)80-10mm管道的檢測(cè)，我們開(kāi)發(fā)了一種新型的多傳感器管道檢測(cè)機(jī)器人。該機(jī)器人裝置由兩個(gè)平面形狀的鏈輪組成。只需使用一個(gè)機(jī)器人模塊即可產(chǎn)生轉(zhuǎn)向和運(yùn)動(dòng)運(yùn)動(dòng)，無(wú)需任何奇異運(yùn)動(dòng)。這個(gè)機(jī)器人的平板形狀允許在機(jī)器人的兩側(cè)安裝額外的傳感器。通過(guò)在試驗(yàn)臺(tái)環(huán)境下的各種實(shí)驗(yàn)，驗(yàn)證了所提出的管道檢測(cè)機(jī)器人系統(tǒng)的性能。 VIII.確認(rèn) 這項(xiàng)工作是部分支持的中期研究項(xiàng)目通過(guò)NRF贈(zèng)款資助的最高明的(2010 - 0000247),部分支持GRRC京畿道計(jì)劃(2010 - a02 GRRC漢陽(yáng)),部分支持的知識(shí)經(jīng)濟(jì)部(MKE)和韓國(guó)發(fā)展研究所技術(shù)(吉)在戰(zhàn)略技術(shù),通過(guò)員工發(fā)展計(jì)劃和部分支持的人力資源開(kāi)發(fā)韓國(guó)能源技術(shù)研究所韓國(guó)知識(shí)經(jīng)濟(jì)部評(píng)估與規(guī)劃(KETEP)資助項(xiàng)目。 參考文獻(xiàn) [1]H.T.Roman, B. A. Pellegrino, and W. R. Sigrist, “Pipe crawling inspection robots: An overview,” IEEE Trans. Energy Convers., vol. 8, pp. 576–583, Sept, 1993. [2]W. Neubauer, “A spider-like robot that climbs vertically in ducts or pipes,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst., 1994, pp. 1178–1185. [3]S. Hirose, H. Ohno, T. Mitsui, and K. Suyama, “Design of in-pipe inspection vehicles for _25, _50, _150 pipes,” in Proc. IEEE Int. Conf. Robot. Autom., 1999, pp. 2309–2314. [4]J. Okamoto, Jr., J. C. Adamowski, M. S. G. Tsuzuki, F. Buiochi, and C.S. Camerini, “Autonomous system for oil pipelines inspection,” Mechatronics, vol. 9, pp. 731–743, 1999. [5]M. M. Moghaddam and A. Hadi, “Control and guidance of a pipe inspection-PIC,” in Proc. Int. Symp. Automations, Robotics, 2005, pp. 11–14. [6]T. Aoki and S. Hirose, “Study on the brake operation of bridle bellows,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst., 2007, pp. 40-45. [7]P. Li, S. Ma, B. Li, and Y. Wang, “Design of a mobile mechanism possessing driving ability and detecting function for in-pipe inspection” in Proc. IEEE Int. Conf. Robot. Autom., 2008, pp. 3992-3997. [8]M. M. Moghaddam and A. Hadi, “Control and guidance of a pipe inspection-PIC,” in Proc. Int. Symp. Automations, Robotics, 2005, pp. 11–14. [9]Y. S. Kwon, H. Lim, E. J. Jung, and B.-J. Yi “Design and motion planning of a two-moduled indoor pipeline inspection robot,” in Proc. IEEE Int. Conf. Robot. Autom., 2008, pp. 3998–4004. [10]S. G. Roh, D. W. Kim, J. S. Lee, H. P. Moon, and H. R. Choi, “Modularized in-pipe robot capable of selective navigation inside of pipelines,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst., 2008, pp. 1724–1729. [11]Y. S. Kwon and B.-J. Yi, “The kinematic modeling and optimal paramerization of an omni-directional pipeline robot," in Proc. IEEE Int. Conf. Robot. Autom., 2009, pp. 1389-1394. [12]C. Anthierens, C. Libersa, M. Touaibia, M. Betemps, M. Arsicault, and N. Chaillet, “Micro robots dedicated to small diameter canalization exploration,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst., 2000, pp. 480–485. [13]S. G. Roh, S. M. Ryew, J. H. Yang, and H. R. Choi, “Actively steerable inpipe inspection robots for underground urban gas pipelines,” in Proc. IEEE Int. Conf. Robot. Autom., 2001, pp. 761–766. [14]T. Okada and T. Kanade, “A three-wheeled self-adjusting vehicle in a pipe, FERRET-1,” Int. J. Robot. Res., vol. 6, no. 4, pp. 60–75, 1987. [15]S. Fujiwara, R. Kanehara, T. Okada, and T. Sanemori, “An articulated multi-vehicle robot for inspection and testing of pipeline interiors,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst., 1993, pp. 509–516. [16]T. Oya and T. Okada, “Development of a steerable, wheel-type, in-pipe robot and its path planning,” Advanced Robotics, Vol. 19, No. 6, pp. 635-650, 2005 A Flat Pipeline Inspection Robot with Two Wheel Chains ABSTRACT This paper presents a new pipeline inspection robot that has multiple sensors for inspection of 80-100mm pipelines. The special feature of this robot is realization of driving and steering capability by using only two wheel chains. Compared to popularly employed pipeline robots using three wheel chains, the new design allows simple robot control and easy user interface, specially at T-branch. As another advantage, the flat shape of this robot allows mounting additional sensors on the both sides of the robot. The kinematics and three control modes are described. Finally, the performance of this robot system is verified by experimentation. Keywords：Pipeline robo；system；kinematics I.INTRODUCTION The functions of pipeline inspection robots can be described as driving, steering, detecting, and retrieving. However, robots used for inspection of pipelines with its diameter of less than 100mm has a special difficulty in designing compact steering mechanism and mounting sensors such as magnetic probe and ultra-sonic probe to inspect crack, rupture, leaking, etc. In-pipe robot mechanism, which has a long history of development in robotics, can be classified into several elementary forms according to the movement patterns. They are wheel -type, inchworm-type, legged mobile-type, screw-type, crawler-type, PIG-type, and passive -type. Among them, wheel-type pipeline inspection robots have been mostly popular [1]-[8]. However, they are not appropriate for operations in the vertical pathway or at T-branch. During the latest 10 years, differential-drive type mechanisms have been studied intensively [9]-[11]. The differential-drive type usually has three powered wheel chains. Controlling the speed of each chain independently, the robot is able to go through elbows and T-branches. However, when using just one robot module, sometimes singular motion happens at T- branch[9] . To resolve such a problem, several methods have been developed such as active steering joint mechanism [12-13] or collaboration of two robot modules[9]. However, the size of the whole robot system becomes bulky. Another disadvantage of using three powered wheel chains is not having enough space to mount more sensors in the robot body, because three wheel chains occupy most of cross-sectional area of the pipeline specially having diameter of less than 100mm. Currently, only one camera is installed in front of the robot body. T. Okada, et al[14-16] developed flat type pipeline inspection robot. However, their design is complex and used for large sized pipelines. In light of these factors, we propose a pipeline inspection mechanism having two powered wheel chains. The two wheel chains are arranged with 180 degrees apart, so additional sensors can be attached on both sides of the robot body. Each wheel chain is controlled by using two motors ; one for driving and the other for steering. A set of count wheels creates a screw motion to realize steering and another set of wheels creates a linear motion inside the pipeline without having any singular motion at elbow or T-branch. The robot system consists of 3 control modes ; moving, detecting, and searching. We will introduce corresponding strategy for each mode. The robot mechanism is designed using a parallel linkage so that it can provide the foldable characteristic. The foldable characteristic allows adaptation of the wheel mechanism to the wall inside the pipeline. Section II introduces the characteristics of the robot system. The kinematic analysis is presented in section III. We show the validity of this robot system by both simulation and experimentation in section IV. Lastly, we draw conclusion. II.CHARACTERISTICS OF ROBOT A.The whole system and robot device The robot system shown in the Fig. 1 consists of a control box and a robot device. The robot system is controlled by a joystick interface. The user of robot can check the state of pipeline by using vision information. The robot device consists of a main body and two wheel chains, as shown in Fig. 2. The length of robot is 80mm and the exterior diameter is 100mm. Fig. 1. The pipeline inspection robot system with two wheel chain mechanism. This robot mechanism can be operated in three different modes; driving mode, detecting mode, and searching mode. In the driving mode, the robot is in motion. In the detecting mode, the robot detects the state of the inner wall of the pipeline. In the searching mode, the robot collects the detail information. （a） （b） Fig. 2. The whole structure of the pipeline inspection robot system with two wheel chain mechanism: (a) The 3D model, (b) The developed robot. B. The main body and side sensor mounter Fig. 3 shows that the main body consists of a key slider, two main body axes, four compression springs, a sensor mounter, a front camera, and two side sensors. The key slider(prismatic joint) slides on the main body axis and it is connected to the linkage of the wheel chain. The compression spring adapts to change of the outer diameter of the wheel chain. This design allows a foldable characteristic of the robot body. Each key slider is connected by a key connector so that two wheels generate the same folding motion. Thus, the main body is always located at the center of the pipeline if we use this design. In this mechanism, the robot has a flat shape by using two chains. Thus, there are some spaces on left and right sides of the robot body, where various sensors like MT, UT, or vision sensor (camera) can be mounted. In this paper, we use a front camera, the right camera, and the left camera as shown in the Fig.3. From each camera, we can get the internal information of the pipeline as shown in Fig. 4. Fig. 3. The structure of the main body Fig. 4. The camera view - front camera, left camera, and right camera C. The wheel chain mechanism The main body is connected to the folding mechanism of the wheel chain. The wheel chain consists of a folding mechanism (parallel linkage), a driving motor (10 ), an active wheel, a steering wheel (idle wheel), a steering motor (8 ), and a steering mechanism as shown in Fig. 2 and Fig. 5.The wheel drive motor drives the active wheel through a helical gear power transmission. As shown in the front view of Fig. 5, the steering motor changes the direction of the steering wheel so that the robot can be steered to a desired direction. Fig. 5. The structure of the wheel chain III.ROBOT MOTION PLANNING AND DETECTING FUNCTION The pipeline inspection robot system is required to pass through straight and curved pipelines. To do this, we need to develop new motion planning algorithms. A. The basic movement at the pipeline Fig. 6. The motion at the straight pipeline (a) the front view (b) the isometric view (c) the top view xyz in Fig. 6(a) represents the local coordinate frameattached to the body of the pipeline inspection robot. We define oi as the steering angle of the steering wheel. If there is no steering angle, the robot moves forward and backward along the z -axis without any rotation in the pipeline. The forward and backward movements of the robot are used in the driving mode when the robot operator checks the state of pipelines by using the front camera. Fig. 7 shows the rotational motion of the robot. If the two steering wheels are arranged in an opposite direction such as Fig. 7(a), the robot rotates clock wisely and creates a screw. If the wheels are arranged in an opposite direction, the robot rotates counter clock wisely. Such screw motion can be used in the detection mode; Once the front camera detects some flaws on the inner surface of the pipeline, the robot uses the screw motion and locate the position of the side camera such that the side camera is able to inspect the spot clearly. Steering motion at elbows or T- branches can be also achieved by controlling the angles of the steering wheels in the same direction. Fig. 8(a) shows that the robot turns to the left at the T-branch, when the two steering wheels are steered to the left direction. Fig. 8(b) is for steering to the right direction. It should be noted that the speeds of the two driving motors are the same even in the steering motion. Thus, control effort for steering is less as compared to three powered wheel robot in which three wheels have differential wheel speed to pass through elbows or T-branches. Fig. 8. The motion at the T-branch and elbow (a)Left turning motion (b)Right turning motion. B.The control modes The pipeline inspection robot under investigation has control modes; driving mode, detecting mode, and searching mode. First of all, in the driving mode the robot moves forward or backward to inspect the state of the pipeline by using the front camera. In the detecting mode, the position of the special spot on the wall is exactly located by using both the front and side cameras. In the searching mode, the state or problem of the spot is identified by using the side camera. Fig. 9 shows the pictures of the three modes. Fig. 9. The experimental result : navigation inside the pipeline IV. ANALYSIS OF MECHANISM The detailed structure, coordinate system, joint variables, and parameters of the wheel mechanism are given in Fig. 10. xyz denotes a local coordinate frame attached to the pipeline. Fig. 10. Folding mechanism V．IMPLEMENTATION A. Controller The robot controller consists of a control box, a control PC, a grabber board, a joystick interface, and a robot device as shown in the Fig. 12. The robot control is executed by a serial communication. In this system, we use MCUs (Atmega128). The MCU calculates the motor speed by producing a PWM signal. It can control all of the Micro DC motors. All the motor drives and MCU are integrated in the control box. The view of the pipeline is provided to the user by using a Micro CMOS camera mounted in front of the robot body. This camera module makes it possible to inspect the condition inside the pipeline. The robot device equipped with a camera is shown in Fig. 12. Fig. 12. The robot controller. B. The robot device The motors are embedded in the motor box of the wheel mechanism. The peak torque of steering motor is 30 mNm. The Maxon re 8 and GP 8B gear head are chosen for steering motor. The peak torque of the driving motor is 150 mNm. The Maxon re 10 and GP 10A gear head are chosen for the driving motor. Table I shows the specification of the robot. The length of the robot module is 80mm and the exterior diameter of the robot body changes from80mm up to 100mm. The total length of the robot device including the camera and the lighting device is 94mm. And the weight of the robot is 237g. In this paper, a pipeline with diameter of 100 mm is employed as a test bed. C. The camera In this paper, three CMOS camera sensors were used. Its diameter is 10mm, and its minimum focal length is 2cm such that it can be used in the pipeline with diameter of less than100mm. Table II shows the specification of the camera. TABLE I SPECIFICATION OF THE ROBOT Specification Tbot-100-2chn Weight of the robot module 237g Motor diameter 6mm Length of the robot module 80mm Total length of the robot including camera) 94mm Exterior diameter 80-100mm Linear speed 14cm/sec Serial communication distance 15 M VI. EXPERIMENTAL RESULTS A pipeline consisting of a straight part, a T-branch, and an elbow was employed in experiment. Its interior diameter is 100mm. The T-branch and elbow are commercial products. Fig. 13. The test bed. Fig. 13 shows the test bed for the experimentation and Fig. 14 shows the driving and steering motions at elbow. Fig. 15, 16 demonstrates that the pipeline inspection robot is able to steer at the T-branch and can also change the direction successfully by combining the backward and steering motions. Thus, the performance of the proposed robot system could be verified through this experimentation. The attached video clip shows the experimental result. Fig. 14. Motion planning at the elbow : (a)~(b) forward motion at straight pipeline, (c) wheel steering, (d)~(f) turning to the left direction at elbow VII. CONCLUSIONS We developed a new pipeline inspection robot that has multiple sensors for inspection of 80-10mm pipelines. The robot device consists of two wheel chains which has a flat shape. The steering and moving motion can be generated by using just one robot module without having any singular motion. The plat shape of this robot allows mounting additional sensors on the both sides of the robot. The performance of the proposed pipeline inspection robot system was verified through a variety of experiment under a test-bed environment. ACKNOWLEDGEMENT This work was partially supported by Mid-career Researcher Program through NRF grant funded by the MEST (No. 2010-0000247),partially supported by GRRC program of Gyeonggi Province (GRRC HANYANG 2010