CNC機床或坐標測量機外文文獻翻譯、中英文翻譯、外文翻譯

CNC機床或坐標測量機外文文獻翻譯、中英文翻譯、外文翻譯,CNC,機床,坐標,測量,外文,文獻,翻譯,中英文 附錄一 CNC機床或坐標測量機 這論文描述在CNC機床或坐標測量機（CMM）和它的CAD/CAM綜合為任何的自由形態(tài)的表面發(fā)展一個非連絡類型自動機械測量系統(tǒng)的工作。 一個Keyence公司模型LC-2220激光探針,作為非連絡感應器被整合到CNC機器之內(nèi)。已經(jīng)發(fā)展了一個為任何自由形態(tài)輪廓的自動表面追蹤的測量軟件。對于反向工程任何的市售 CAD/CAM系統(tǒng)的轉換數(shù)據(jù)，通過正確的DXF文件,也是可得的。 廣泛的校準工作，在關于顏色，材料，表面斜坡，和使用HP5528激光干涉儀的工件邊緣偵測方面,激光探測的系統(tǒng)精確方面已經(jīng)實現(xiàn)了。在使用表面描繪技術后，副本物體的形狀誤差，相對于它的樣品，不超過30微米，對于鑄型業(yè)應該是足夠了。 自從1960年，英國的Ferranti公司發(fā)展了第一部坐標測量機(CMM)，空間度量衡學的測定效率，已經(jīng)在工業(yè)中得到很大的改良。照慣例，大部分CMMs裝備有接觸探針，為了在如線，平面，圓，柱面，球，圓錐體，等幾何因數(shù)方面的接觸類型測量。測量的程序在以上因素方面的確非?？於铱芍貜停驗橹恍枰綔y點的少數(shù)數(shù)字，而且探測半徑的補償是相當簡單的。然而當3D自由形態(tài)表面測量的需要近幾年來增加，尤其隨著反向工程的鑄型業(yè)，當在很多測量點方面應用接觸探測時，許多缺點被發(fā)現(xiàn)。一些典型的問題是：由于重復的進出表面的運動， 速度不是夠快。被抽取樣品的位置因探測速度而影響，被抽取樣品的點可能不夠適當來正確地表現(xiàn)標準的表面，探頭頂尖的半徑補償更難。探頭頂尖容易很快磨損，而且一些機械結構和伺服控制器的誤差可能會發(fā)生。 為了改善上述缺點，在過去四年，有些激光探頭已經(jīng)開發(fā)出來，用于自由曲面的非接觸式測量。此外，在應用這種探針到CNC型的CMM時，系統(tǒng)的測量精度和速度可以大大提高5-6倍，及CAD / CAM集成可以很容易地實現(xiàn)其逆向工程的目的。 這論文描述使用了一個低成本激光探針到一個DCC（直接的計算機控制）型的CMM來為3D自由形態(tài)表面和基于反響工程的CAD/CAM軟件發(fā)展一個自動的非接觸測量系統(tǒng)。如此的技術也被適用于 CNC機器中心來建立一個機器測量系統(tǒng)。激光的一個深入口徑測定探查主要是調(diào)查有關的材料，表面坡度，顏色，以及待測 量工件的邊緣檢測 。在使用表面涂層技術后，再生產(chǎn)的工件形狀誤差相對于它的主件在 30jrn 之內(nèi)， 這個誤差對于模具業(yè)的要求應該是足夠的。2激光探針的工作原理。 用于此項研究的激光探針是一個帶有LC–2100的KeyenceLC-2220傳感器頭。這種成本低可見的激光探針通過激光三角提供位移測量技術。正如所示，從半導體發(fā)射器發(fā)出的聚焦的激光束投影到物體表面。由于表面粗糙度，反射光束將分散和通過激光接收鏡頭部分收集。通過這個鏡頭，光束將集中和投影到位置敏感探測器(PSD)的傳感器頭。如果被測量物體移動(Lx,圖中), 反射激光光斑也通過PSD手段移動。然后，這點位移通過這里也被稱為控制器的數(shù)據(jù)處理轉換為一個模擬或數(shù)字信號。最后，該控制器進行數(shù)據(jù)處理，如線性補償和平均值計算，顯示和輸出測量結果。 所有的非接觸式位移傳感器，如渦流型，超聲波型，氣動型，激光束反映類型，它們的性能曲線都與測試對象的表面性質(zhì)有關。一些激光探針表面特性的重要參數(shù)，包括材料，顏色，粗糙度，以及物體表面的斜坡。大多數(shù)激光探測器的規(guī)格，然而，業(yè)務手冊中沒有提供足夠的信息。所列出的反應制造商所提供的任何特定的激光探針的性能特點的數(shù)據(jù)通常沒有直接適用于特定的加以衡量的對象。因此，首要的工件方面的激光探針必須被采用以保證測量結果更準確。 顯示了激光探針標定系統(tǒng)實驗裝置。進行校準的激光探針是安裝在三坐標測量機(CMM, DCC 類型, Numerex 有限公司）。那個測試試樣放在一個正弦把手上。正弦波的角度和高度能被規(guī)區(qū)塊千斤頂分別地調(diào)整。HP5528激光干涉儀用于提供相關相對于激光探針輸出的丟失。主軸可由計算機逐步移至向上和向下。因此， 一旦開始，這個校準任務可根據(jù)德國工程師協(xié)會3441標準自動處理。而且每個操作被運行五次雙方向的行進。各種實驗然后改變不同的如材料（鋼，鋁合金，和電木）斜坡（ 0到60度的步驟） ，顏色（原始的，和白色，紅色，黃色著色）， 和表面粗糙度(0.4pm到3pm)的試樣條件調(diào)查序列。 眾多的校準數(shù)據(jù)是不同的試樣條件方面收集的。說明了研究電木標本方面典型圖位移誤差，其原來的顏色（棕） ，和有激光束的表面。結果發(fā)現(xiàn)，在測量范圍（±2毫米）的最大位移誤差為-l8pm，但重復性和扭轉錯誤都非常好。圖 3 b它的直線性有最少－角尺線彎曲。總結在同樣的條件下五個測試結果。一個有 趣的現(xiàn)象可以看見，表面坡度也增加了位移誤差。它可以解釋說，如圖4所示物體表面上散射光沿鏡面方向執(zhí)行高斯分布。隨著表面正常正在遠離激光束軸，被反映的由感應器頭中的光接受透鏡收集光束的強度會降低,這將導致低信號噪聲(S/N)。 因此，線性度會越來越差。然而，幸運的是，重復性仍然和以前一樣好(在2pm之內(nèi))。 類似的研究也分別在鋁合金樣品鋼樣品執(zhí)行. 此外，激光探針性能的表面顏色的影響通過在被測試的表面上涂不同的顏色來研究.總結了激光探針關于鋁合金在不同的斜度和不同的表面顏色方面的校準結果. 一個重要現(xiàn)象是,無論表面坡度的變化,紅表面總是可以確保激光探針有很好的精度性能。由于鋁合金試樣改為鋼試樣,類似的結論如所示仍然可以得到.從這些研究中，模具制造行業(yè)可以考慮, 當利用帶有激光探針逆向工程技術生產(chǎn)模具, 如果是涂有紅色油漆,可以大大提高主件的測量精度. 在激光探針測量時,工件表面粗糙度的影響已經(jīng)先前調(diào)查了. 結果發(fā)現(xiàn)，在任何加工表面合理Ra值(0.4to3pm),只要光學平面激光探頭垂直于表面,校準結果并沒有表現(xiàn)出顯著的變化.由于大多數(shù)測量的工件必須擦亮到確定的小的Ra 值，這種影響，因此可以忽略不計。大多數(shù)市場上的激光探針只提供位移測量的功能. 雖然這一功能可以使表面描測量的技術成為可能,但大多數(shù)激光探針在邊緣檢測方面的能力現(xiàn)在仍然非常差. 一些在這一技術的解決方案將會發(fā)生, 因為任何幾何形狀的工件必須有邊界。一種這方面的方法在這項工作中被提議. 所調(diào)查的激光探針，帶有LC - 2100控制器傳感器頭的Keyence LC-222,具有數(shù)字化,模擬輸出端口。數(shù)字端口通過RS - 232C接口或GPIB接口為外部設備傳輸數(shù)字讀數(shù). 同時，類似端口按比例發(fā)送一個電壓值送到儀表閱讀向外到一外部的A/D轉換器。隨著激光探頭是在其測量范圍之外,無效的地區(qū),數(shù)字讀出(DRO）會出現(xiàn)一個“黑色”的模擬信號而且類似端口將輸出6.55伏特。隨著激光探針位于其有效的區(qū)域,在對于表面的參考距離, 模擬信號將輸出電壓零。 最初的實驗在CMM上單位工作表面上的參考距離通過設置激光探針進行的. 然后水平地而且重復地被移動進移動出表面,如圖5所示.模擬信號的電壓變化通過數(shù)字示波器觀察到.顯示了一個典型展示,當通過工作表面的銳利的邊緣的電壓變化.清楚地看到,當沿銳利邊緣從無效區(qū)域到有效的區(qū)域,輸出信號出現(xiàn)明顯 和進給速度成正比的延遲時間.然而，當從有效區(qū)域到無效區(qū)域移動探針時,在邊緣地區(qū)輸出信號立即作出反應.這種現(xiàn)象是很難解釋。然而作者認為，這可能是由于激光探針的三角原理。然而，指出了一個可行的解決辦法，該邊緣檢測應該從有效的區(qū)域到無效的區(qū)域做到.另一個重要因素還應當指出是,光學平面的運動必須垂直于表面位置,否則光將嚴重分散。為了驗證激光探針中的邊緣檢測的增強的能力,三維測量在一條線和一個圓分別測量. 顯示每個被提議的測量路徑. 圖8A和圖8B測量結果都非常符合相應的標稱尺寸。 對被提議的工具路徑傳用和控制方法的概要描述是描述在圖7中。在被提議的方法中，合成物剪裁者聯(lián)絡數(shù)據(jù)，包括CC位置和表面原則常態(tài)（N）的云形規(guī)功能，預先設定CC速度（Vc ），和工具傾向及傾斜角度（Φ和λ），與CNC機器聯(lián)系。在CNC內(nèi)插器中，CL路徑是被計算為在真正的時間內(nèi)沿著CC路徑所需要的CC速度。綜上，彌補運算法則的刀具是在CNC內(nèi)插器中實現(xiàn)的。 一個環(huán)型的反饋圈被內(nèi)插器所封閉，這個反饋圈能檢測到在真正時間里的實際的CC位置（ C* ），以及需償還的偏離指定CC路徑的錯誤。除此之外，在線的改編的同盟（i.e CC速度）是介紹增加機制準確性和效率。舉例來說，當CC偏離錯誤大量增加時我們可以降低機制（通過降低Vc ）以及當偏離錯誤可以忽略時加速機制（通過減少Vc ）。注意一些變數(shù)同盟（適當?shù)卦诰€逐漸增加減退的Vc ） 僅僅在它在其他同盟-受扶者切斷效果時不引起無法接受的降格時被保留，這些 切斷效果就是像切斷時間，工具包裝，表面正直，等等。對被提議的工具路徑傳 用和控制方法需要的運算法則被呈現(xiàn)在下列各項。 CC 的路徑竄改，剪裁彌補和倒轉的運動學轉變給內(nèi)插器的核心功能是CC路徑竄改，剪裁彌補和倒轉的運動學的轉變。讓為CC位置（C）和表面原則常態(tài)（N） 所提供的云行規(guī)功能是通過以下公式表述出來的。所用的u是云行規(guī)向前空間的參數(shù)。 在上段決定了的竄改運算法則是在這里應用。因為這個竄改法則是引導CC 的輸出路徑而非CL的路徑。為符合CC路徑，公式中的t需要用c來替換。此外，在式中的CL路徑的速度（Vl ）需要被CC路徑的速度（Vc ）來替換。 在式彌補運算法則的刀具對及時的刀補是適當?shù)摹Ｔ谑阶?B = T ′ N 中的沿 著CC路徑的單位矢量T可以通過下面的式子計算出來通過刀具補償，刀具位置和定方向是可得的。根據(jù)式子（5c）的描述，基于倒轉的運動學是可轉變的，我們把參考指令給5軸的動軸，R。 附加到上方的核心功能的還有，兩個可選的功能包括，一個是在真正的時間里CC路徑的偏離錯誤，另一個是被介紹為改變CC的速度來增加控制表現(xiàn)的。這兩個功能都是基于實際CC點的，即C* 。在典型的5軸工作母機運動控制中，可得的 回應數(shù)據(jù)是沿著這5個軸的輸出位置?；谠冢?a）和（4b）中提及的運動學轉變原理，我們依下列各項可以得到實際的剪裁位置（ L* , O* )。 根據(jù)式可以計算出實際CC點C* 對應的是（ L* , O* ）練習倒轉的刀具彌補。 因為它包括實際傾向的計算和傾斜角度（f* 和l* ）的定義有關于TNB的框架，所 以倒轉的形成非常復雜。在下列各項中，提議了一個簡單但是有效的運算法則。從幾何學角度看，CC錯誤， C* - C 是CL錯誤的重疊， L* - L ，以及刀具定位錯誤 O* - O 。設Ψ偏離刀具軸，U是沿著Ψ的旋轉方向的單位矢量。因為CC點是一個固定在刀具上的點，我們有式中 注意式僅適用于當Ψ與l比足夠小的時候。然而這種情況對常態(tài)的5軸的工作 母機運動控制總是真實的。（注意在以后的模擬例子中，對于Ψ的典型價值是少于0.005rad）。 可以看出，CC路徑偏離錯誤，δ，是指從C* 到指定的表面的距離，這個是我們關注的部分，因為它直接決定機制錯誤。偏離錯誤可以接近地表示為 注意偏離錯誤的方向是在表面常態(tài)，也就是δ=δN。 在抽取樣品kh中分別是參考CC位置的初始和償還值。 在實際運動控制中除去或者補償偏離錯誤，在抽取樣品時參考CC位置可以用下式來修正 k k 式中C 和Ccomp Kp 是比 例控制增益的補償環(huán)。注意其他控制規(guī)律（舉例來說比例的控制，領引-落后的控制，等等）可以被介紹為被第二段的所替代。當然，在這里一些簡單的比例控制也是需要考慮在內(nèi)的。控制得到， Kp ，必須是能夠有效地減少δ和維持系統(tǒng) 的穩(wěn)定。在后面的說明例子中，Ziegler-Nichols 終極-敏感的方法[23]被利用來決定 Kp 回饋圈的另一個作用就是收益一個沿著CC路徑的變數(shù)進給以此來改善機制的準確性和效率。在此期間，我們可以減少Vc 作為δ的增加，這個方法就是減慢機制就像藉由δ的表示改善機制的準確性一樣。另一方面，我們增加Vc 可以以δ 為基礎，我們有Vc =Vc (δ)。在實踐中，僅僅當機制增加時間是允許的時候減 速是能夠被獲得。此外，僅僅當加速在其他進給-依賴切斷效果（像工具外表， 表面正直，等等）不引起無法接受的降格時才考慮它。在后面說明的例子中，在線的同盟改編焦點是改良機制的準確性（也就是減少δ）。 在下面的示范中，一個規(guī)定的表面是生產(chǎn)5軸的機制。規(guī)定表面有下列各項參數(shù)的形式 提及了工作母機的幾何學參數(shù)是( lx , ly , lz )=(0,0,500)毫米和bz =200毫米。最 后利用rt = 5毫米和rc = 0毫米。那需要的是進給是10mm/s。CC路徑是預設在u方 向上而路徑間隔v方向。工具軸分配為Ф=20°和λ=0（注意傾向角度Φ一定比17.9°大這樣才能避免后面凹的范圍的精確計量）。顯示了加工過的表面和一條特殊的CC路徑，C(u)，沿用v=0和它對應的CL路徑，L(u) 和O(u)。 在傳統(tǒng)的方法中，CC數(shù)據(jù)，L（u）和O(u)連同CL速度Vl ，都被輸入到CNC內(nèi)插器中用以及時的剪裁。注意此處CL的速度Vl 作為進給速度設定為10mm/s?；谏厦娴臄?shù)據(jù)以及在式提到的內(nèi)插器運算法則，剪裁路徑是能產(chǎn)生在線機制的。然而，當做CC速度Vc 檢查時，我們能夠發(fā)現(xiàn)沿著刀具路徑Vc 的改變（Vc 的變化超過了10%）。變量Vc 導致非固定機制的質(zhì)量和效率。 和上面的問題結合，被提議的工具路徑傳用方法，可以被介紹為決定上面的 區(qū)段。在被提議的方案中，CC數(shù)據(jù)，C（u）和N(u)，連同被設定為10mm/s的CC 速度，全部被輸入到CNC內(nèi)插器中。因為工具路徑的傳用是基于一個固定的Vc ， 一個常數(shù)沿著CC路徑是容易被控制的。 為了要了解5軸的機制，機制工具同時受到驅(qū)策3滑和兩個回轉的軸。每個軸 的運動是根據(jù)伺服促使電機運轉和軸控制器的控制。在下面的模擬方面，數(shù)學型號（在s領域）的選擇[23]是根據(jù)電機受到驅(qū)策的伺服機制 式中i=x,y,z,а,в。在上面的等式中，si 和ti 分別是每個軸的驅(qū)動的增益和時間常數(shù)。si 的值和ti 在模擬方面的應用列在。 系統(tǒng)參數(shù)在模擬方面的應用樣本抽取時期: h=0.002 s 軸的增益和時間常數(shù)（s） 典型的比例-整體-引出的控制規(guī)律是被軸的控制器所利用?？刂破鞯囊苿庸δ埽▃-領域）被[22]所描述 式中 H p ， Hi ， Hd 分別地是比例，整體和引出增益。 H p ， Hi ， Hd 的值在表格1中可以計算出來。 為了達到有效的CC路徑控制，為CC偏離錯誤構造一個環(huán)型的回饋圈，用δ作補償?；谑?jīng)Q定的系統(tǒng)參數(shù)和運算法則，由此可見，被提議的補償方法可以有效地減少δ。此外， 錯誤減少的因素和償還成比例增益， Kp .( Kp >8)。Ziegler-Nichols終極-敏感方法大約是4，哪一個符合就減少一個5的δ。 一個及時的進給運算法則是家少改善CC路徑控制的準確性的。用來旋轉CC 的速度的適當方法比如適合的控制和模糊邏輯控制，這些在下面有所描述。 If dk If dk > dd ,V k =hV k -1 c c c c < dr ,V k = xV k -1 (deceleration) (recovery) c c c c If V k > V o ,V k = V o  (maximum or preset speed) V k < V ,V k = V If c min c min (minimum speed) V k 在上面的規(guī)則中， c 和V k -1 分別是kh 和(k-1)h的速度指令，δk是在時間 c kh 中有計劃的路徑偏離錯誤。η和ξ分別是句頂速度減速的比率和恢復的因素。 V V V o V c 是在 min 和 c 范圍之內(nèi)的。在實踐中，預先設定CC的速度是 c (=10mm/s)。 在 下 面 的 模 擬 中 ， 我 們 設 置 這 些 參 數(shù) 是 ： η =0.99, ξ =1.01, dd =0.01mm,dr =0.0075mm, Vmin =5 mm/s?；谏厦娴乃俣雀木庍\算法則，模擬5 軸運動，結果在兔13中顯示。隨著這些提議的方法，偏離錯誤在速度改編區(qū)域期間是減少的。然而，機制時間增加了0.1s。 對于5軸的機制申請，我們主要關心的部分是在刀具聯(lián)絡速度和沿著刀具方向除去偏離錯誤聯(lián)絡路徑上，而不是刀具位置速度和傳統(tǒng)刀具控制路徑和方法上。為了處理這些擔心，一種新的方法呈現(xiàn)出來，就是及時的刀具路徑傳用和控制。 被提議的刀具路徑傳用方法是刀具-聯(lián)絡路徑的竄改，刀補包含及時的運算法則，和同等的關系轉變。隨著這些方法的提出，刀具路徑傳用以此來滿足沿著剪裁路徑的需要速度對表面的加工。 為了改善在實際5軸的機制中的受約束的準確性，構造另一個環(huán)型回饋圈用CNC內(nèi)插器關閉。環(huán)型反饋中，被關注的沿著刀具路徑的偏離錯誤在在線機制中可以計算出來?；诃h(huán)型回饋圈，補償運算法則介紹的是直接和有效地去除偏離錯誤。此外，如果機制時間的增加是允許的，我們可以得到一個進給的改編運算法則，當偏離錯誤太大是這個運算法則可以用來減少速度。模擬結果已經(jīng)顯示錯誤補償方法的效力和適當?shù)倪M給改編方法。然而，未來在運算法則上的研究或者規(guī)則錯誤補償和進給的改編是推薦的。 附錄二  CNC machine toolor CMM This paper describes the work to develop a non-contact type automatic measurement system for any free-form surfaces on a CNC machine tool or a coordinate measuring machine (CMM), and its CAD/CAM integration. A laser probe, made by Keyence Co. model LC-2220, was integrated into the CNC machine as the non-contact sensor. A measurement software has been developed for automatic surface tracing of any free-form profile. Data transfer to any commercially available CAD/CAM system for reverse engineering is also available via proper DXF file. Extensive calibration work has been carried out on the systematic accuracy of the laser probe with respect to the color, material, surface slope, and edge detection of the workpiece by the use of a HP5528 laser interferometer system. Having employed the surface painting technique, the shape error of the copied object relative to its master piece was found within 30 micrometers, which is deemed adequate enough to the mold industry. Since the first coordinate measuring machine (CMM) was developed by the Ferranti company of UK in 1960, the measuring efficiency of dimensional metrology has been greatly improved in industry. Conventionally, most CMMs are equipped with the touch-trigger probes for contact type of measurement on geometrical elements, such as line, plane, circle, cylinder, sphere, cone, etc. The measuring process is indeed very fast and repeatable with respect to the above elements since it needs only a limited number of probing points and the probe radius compensation is quite straightforward. However, as the demands for 3D free-form surface measurements are increasing in recent years, especially by the mold industry for reverse engineering, many disadvantages have been discovered when applying the contact probe for numerous measuring points. Some typical problems are: the speed is not fast enough due to repetitive motion into and out of the surface, the sampled position is affected by the probing speed, the sampled points may not be adequate enough to represent the measured surface accurately, the technique of probe radius compensation is more difficult,\the probe tip is subject to be worn quickly, and\some dynamic errors of the machine structure and the servo controller may occur. In order to improve the above-mentioned drawbacks, some laser probes have been developed for non-contact measurement of free-form surfaces during the past years'4. In addition, when applying this kind of probe to the CNC type of CMM, the system accuracy and the measuring speed can be significantly increased5'6, and the CAD/CAM integration can be easily achieved for the purpose of reverse engineering7'8. This paper describes the work which employed a low cost laser probe to a DCC (Direct Computer Controlled) type CMM to develop an automatic non-contact measuring system for 3D free-form surfaces, and its integration with some PC based CAD/CAM softwares for reverse engineering. Such technique was also applied to a CNC machining center to build up an on-machine measurement system. An in-depth calibration of the laser probe was primarily investigated with respect to the material, the surface slope, the color, and the edge detection of the workpiece to be measured. Having employed the surface painting technique, the shape error of the reproduced workpiece relative to its master piece was found within 30 jrn, which is deemed adequate enough to the requirement by the mold industry. The laser probe used in this study is a Keyence LC-2220 sensor head with LC-2100 controller. This low cost and visible laser probe provides displacement measurement via laser triangulation technique. As seen in Fig. 1, the focused laser beam emitted from the semiconductor laser is projected onto the object surface. Due to the surface roughness, the reflected beam will be scattered and partly collected by the laser receiving lens. Through this lens, the beam will be focused and projected onto the position sensitive detector (PSD) in the sensor head. If the object to be measured moves (Lx, in the figure), the reflected laser light spot also moves (6) by means of the PSD. Then, this spot displacement is converted to an analog or a digital signal through a data processing unit which is also called the controller here. Finally, the controller carries out data processing, such as linearity compensation and average value calculation, displaying and outputting the measured results. For all of the non-contact type displacement sensors, such as the eddy current type, the ultrasonic type, the pneumatic type, and the laser beam reflected type, their performance curves are related to the surface properties of the tested objects. Some important parameters of the surface property with respect to the laser probe may include the material, the color, the roughness, and the slope of the object surface. Most of the specifications of laser probes, however, do not provide sufficient information in the operational manuals. The listed data reflecting the performace characteristics of any particular laser probe provided by the manufacturer uaually are not directly applicable to a particular object to be measured. Therefore, primary accuracy calibration of the adopted laser probe with respect to the adopted workpiece has to be carried out in order to guarantee the measured results more accurately. Fig. 2 shows the experimental set-up for this laser probe calibration system. The laser probe to be calibrated was mounted on the spindle head of a coordinate measuring machine (CMM, DCC type, Numerex Co.). The tested specimen was placed on a sine bar. The angle and the elevation of the sine bar can be adjusted by the gage blocks and the jack respectively. A HP5528 laser interferometer was adopted to provide reference displacement in comparison with the laser probe output. The spindle could be moved up and down step-by-step by the computer commands. Therefore, once initiated, this calibration task could be processed automatically according to the VDI 3441 standard° , and five times bi-directional travels were executed for each task. Various experiments were then investigated in sequence by changing different specimen conditions, such as materials (steel, aluminum alloy, and bakelite), slopes (0 to 60 degrees in steps), colors (original, and white, red, yellow paintings), and surface roughnesses (0.4 pm to 3 pm). Numerous calibrated data were collected with respect to different specimen conditions. Fig. 3a illustrates a typical diagram of displacement errors of the investigated laser probe with respect to the bakelite specimen in its original color (brown) , and with its surface normal to the laser beam. It was found that within the measuring range (±2mm) the maximum displacement error was —l8pm, but the repeatability and the reversal error were both very good. Fig. 3b plots its linearity curve with respect to the least-squares line. Table 1 summarizes five tested results at similar specimen condition except with different surface slopes. An interesting phenomenon can be seen that as the surface slope increases the displacement error increases too. It can be explained that the scattered light on the object surface performs a Gaussian distribution along the specular direction'°, as shown in Fig. 4. As the surface normal is moving away from the laser beam axis, the intensity of the reflected beam collected by the light receiving lens in the sensor head is decreasing which will result in low signal-to-noise ratio (S/N). Consequently, the linearity will get worse. However, fortunately, the repeatability still remains as good as before (within 2pm). Similar studies were also carried out with respect to an aluminum alloy specimen and a steel specimen separately. In addition, the influence of surface colors on the laser probe's performance was studied by uniformly painting different colors on the tested surface. Table 2 summarizes calibrated results of the laser probe with respect to the aluminum alloy at different slopes and in different surface colors. A significant phenomenon was found here that, regardless of the changes in the surface slope, the red surface could always ensure the laser probe with very good accuracy performance. As the aluminum alloy specimen was replaced by the steel specimen, similar conclusion could still be obtained, as seen in Table 3. From these studies, a suggestion can be given to the mold making industry that, while producing the mold by the reverse engineering technique with a laser probe, the measuring accuracy can be significantly improved if the master piece is coated with red paint. The influence of the surface roughness of the workpiece on the laser probe measurement had already been investigated previously". It was found that within the reasonable Ra values (0.4 to 3 pm) of any machined surface, the calibrated results did not show significant changes as long as the optical plane of the laser probe was perpendicular to the surface lay. Since most of the workpieces to be measured must have been polished to certain small Ra values, this effect could therefore be neglected. Most of the laser probes on the market only provide the function of displacement measurement. Although this function can make the technique of surface scanning measurement possible, the capability of most laser probes in the edge detection is still very poor nowadays. Some solutions, in this technology, will have to come about since any geometrical shape of the workpiece must have boundary. A method for this aspect is therefore proposed in this work. The investigated laser probe, Keyence LC-2220 sensor head with LC-2100 controller, has both digital and analog output ports. The digital port transmits its digital readouts to an external device via a RS-232C or a GPIB interface. Meanwhile, the analog port sends a voltage value in proportion to the meter reading out to an external A/D converter. As the laser probe is ou of its measuring range, the invalid region, the digital readout (DRO) will appear a "DARK" signal and the analog port will output 6.55 volts. As the laser probe is located within its valid region and at its reference distance to the surface, the analog signal will output zero volt. An initial experiment was carried out by setting the laser probe, on the 0MM, at its reference distance to a flat work surface. It was then moved in and out of the surface horizontally and repeatedly, as seen in Fig. 5. The voltage changes of the analog signals were observed by a digital oscilloscope. Fig. 6 shows a typical display of the voltage changes when crossing the sharp edge of the work surface. It was clearly seen that when moving the probe from the invalid region into the valid region across the sharp edge the output signal appeared an apparent time delay which was proportional to the feedrate. However, when moving the probe from the valid region to the invalid region the output signal responded immediately at the edge position. This phenomenon is difficult to explain. Yet, the authors believe that it could be due to the triangulation principle of the laser probe. It, however, points out a feasible solution that the edge detection should always be done from the valid region to the invalid region. Another important factor should also be pointed out here that, during motion the optical plane must be perpendicular to the surface lay, otherwise the light will be seriously scattered out. To verify this enhanced capability of the laser probe in the edge detection, dimensional measurements were carried out on a line and a circle separately. Fig. 7a and Fig. 7b show each proposed measuring path. Fig. 8a and Fig. 8b plot the measured results which are all quite consistent with the corresponding nominal dimensions. A schematic description for proposed tool path generation and control method is depicted in Fig.7.In the proposed method., composite cutter contact data, which include the spline functions for the CC location (C) and the surface principle normal (N), the preset CC velocity ( Vc ),and the tool inclination and tilt angles(Φandλ),are loaded to the CNC machine .In the CNC interpolator, the CL path is calculated in real time so as to meet the desired CC velocity along the CC path. Consequently, the cutter offsetting algorithm is implemented in the CNC interpolator. As is shown in Fig.7, a global feedback loop is closed by the interpolator, which can monitor the practical CC location ( C* ) in real time and compensate for the deviation error from the desired CC path. In addition, on-line adaptation of the federate (i.e the CC velocity) can be introduced