外文翻譯--基于智能汽車的智能控制研究

1、<p><b>　　中文4857字</b></p><p>　　畢業(yè)設(shè)計(論文)外文資料翻譯</p><p>　　學(xué) 院： 信息工程學(xué)院 </p><p>　　專 業(yè)： 通信工程 </p><p>　

2、　姓 名： </p><p>　　學(xué) 號： </p><p>　　外文出處： College of Information Science and </p><p>　　Engineering Northeastern Unive

3、rsity </p><p>　　附 件： 1.外文資料翻譯譯文；2.外文原文。 </p><p>　　附件1：外文資料翻譯譯文</p><p>　　基于智能汽車的智能控制研究</p><p>　　摘要：本文使用一個叫做“智能汽車”的平臺進行智能控制研究，該小車采用飛思卡爾半導(dǎo)體

4、公司制造的MC9S12DG128芯片作為主要的控制單元，同時介紹了最小的智能控制系統(tǒng)的設(shè)計和實現(xiàn)智能車的自我追蹤駕駛使用路徑識別算法。智能控制智能車的研究包括:提取路徑信息,自我跟蹤算法實現(xiàn)和方向和速度控制。下文介紹了系統(tǒng)中不同模塊的各自實現(xiàn)功能，最重要部分是智能車的過程智能控制：開環(huán)控制和閉環(huán)控制的應(yīng)用程序包括增量式PID控制算法和魯棒控制算法。最后一步是基于智能控制系統(tǒng)的智能測試。</p><p>　　關(guān)鍵詞

5、：MC9S12DG128；智能控制;開環(huán)控制；PID；魯棒；</p><p><b>　　1.背景介紹</b></p><p>　　隨著控制理論的提高以及信息技術(shù)的快速發(fā)展,智能控制在我們的社會中發(fā)揮著越來越重要的作用。由于嵌入式設(shè)備有小尺寸、低功耗、功能強大等優(yōu)點，相信在這個領(lǐng)域?qū)幸粋€相對廣泛的應(yīng)用,如汽車電子、航空航天、智能家居。如果這些技術(shù)一起工作，它將會蔓

6、延到其他領(lǐng)域。為了研究嵌入式智能控制技術(shù)，“智能汽車”被選為研究平臺,并把MC9S12DG128芯片作為主控單元。通過智能控制,智能汽車可以自主移動,同時跟蹤的路徑。</p><p>　　首先,本文給讀者一個總體介紹智能車輛系統(tǒng)的[2、3]。然后,根據(jù)智能車輛的智能控制:提取路徑信息,自我跟蹤算法實現(xiàn)中,舵機的方向和速度的控制。它提供包括了上述四個方面的細(xì)節(jié)的智能車系統(tǒng)信息。此外,本文強調(diào)了智能車的控制過程應(yīng)用程

7、序包括開環(huán)控制、閉環(huán)增量PID算法和魯棒算法。</p><p>　　2.智能車系統(tǒng)的總體設(shè)計</p><p>　　該系統(tǒng)采用MC9S12DG128[4]作為主芯片，以及一個CCD傳感器作為交通信息收集的傳感器。速度傳感器是基于無線電型光電管的原理開發(fā)。路徑可以CCD傳感器后繪制收集的數(shù)據(jù)，并且系統(tǒng)計算出相應(yīng)的處理。在同時，用由電動馬達速度測試模塊測量的智能汽車的當(dāng)前速度進行響應(yīng)的系統(tǒng)。最后

8、，路徑識別系統(tǒng)利用所述路徑信息和當(dāng)前的速度，以使智能汽車在不同的道路條件的最高速度運行。圖1示出了智能車輛系統(tǒng)的框圖。</p><p>　　3、跟蹤算法的自我實現(xiàn)</p><p>　　智能汽車的自我控制基于其上由CCD傳感器[5]收集的路徑信息。CCD傳感器的數(shù)據(jù)采集速率為25幀/秒。一個幀被劃分為兩個部分：奇數(shù)場和偶數(shù)場，也就是說，50場/秒。為了使該電流路徑的準(zhǔn)確判斷？我們放大頻率為3

9、2MHz。最后，該系統(tǒng)將得到后MC9S12DG128 ATD模塊逆向其由CCD傳感器收集的數(shù)據(jù)35 *307陣列，并忽略一些消隱數(shù)據(jù)。</p><p>　　智能汽車有根據(jù)的滯后運行時，為了提高速度，收集前瞻性路徑信息變得尤為重要?；谏鲜隼碛桑覀冞x擇某一行遠(yuǎn)程數(shù)據(jù)作為智能車控制基本數(shù)據(jù)陣列的第一行。相應(yīng)地，第150行和第300行被選擇作為輔助數(shù)據(jù)，然后我們得到一個新的3* 35二維陣列。在陣列的基礎(chǔ)上，我們可以

10、通過大量的實驗和測試轉(zhuǎn)彎最佳值適應(yīng)的每293行的最佳速度。對應(yīng)的速度和轉(zhuǎn)彎最佳曲線是基于上述數(shù)據(jù)繪制。速度曲線和轉(zhuǎn)彎曲線的擬合過程將使用本文中的第1行數(shù)據(jù)，例如聲明。</p><p>　　圖2示出了智能車與路徑之間的位置關(guān)系（精確地黑色軌道）當(dāng)車轉(zhuǎn)身時L表示該CCD傳感器收集的最遠(yuǎn)基準(zhǔn)線與汽車之間的距離。S表示最遠(yuǎn)中間點的基準(zhǔn)線和所述軌道線之間的距離。R表示的轉(zhuǎn)彎半徑。</p><p>&

11、lt;b>　　計算R的過程如下：</b></p><p>　　從三角函數(shù)推導(dǎo)，我們制定出：</p><p>　　通過測量，L= 0.6米，在90度角范圍內(nèi)，黑色磁道位置和車體位置的垂直中心線之間的距離為約0.2667米。這意味著S的范圍從0到0.2667米。</p><p><b>　　根據(jù)向心力式：</b></p>

12、;<p>　　極限速度可從公式推導(dǎo)：</p><p><b>　?。?）</b></p><p>　　智能車進入角時可以用下式計算出每個基準(zhǔn)點的速度，同時該速度也是最大值。為了計算速度V，我們需要測量的摩擦參數(shù)。</p><p>　　在本文中，計算摩擦參數(shù)的方法如下：首先，智能車放置在KT板上，然后將板的一側(cè)被升高直到智能車可以從

13、板滑動。測量是在地面與KT板之間的角度。其結(jié)果是大約60°該圖3示出了智能汽車的機械分析</p><p>　　以下等式可以得到車體狀況力量的平衡：</p><p>　　mgsin60 = mgsin30</p><p>　　KT板的摩擦系數(shù)，可以計算，u=0.577</p><p>　　該智能車識別圖像陣列的每一行由35個點組成。從

14、左至右，參考點是從1編號到35。在該論文中，0到17點被選擇作為例子進行申報。這些點的最大速度被計算，并且這些數(shù)據(jù)將被顯示在表中</p><p><b>　　轉(zhuǎn)彎的半徑和速度表</b></p><p>　　如可從表中可以看出，智能汽車的最大速度為2.1596米/秒。經(jīng)過測試，結(jié)果表明，速度是2.0米/秒時，PWM等于2400。當(dāng)智能車從一條直線進入彎道前，速度可能比最

15、大速度更快。與此同時，慣性因素作用于智能車。該系統(tǒng)通過了許多次基于最大值的實驗，最后，在表格4中的基準(zhǔn)點的速度曲線是基于從許多實驗中測試得到的實驗數(shù)據(jù)。</p><p><b>　　彎道曲線擬合</b></p><p>　　智能小車方向由轉(zhuǎn)向器的控制。舵機的轉(zhuǎn)向范圍從-45°到+ 45°。</p><p>　　在本文中，MC

16、U總線頻率被PLL乘到32MHz，20ms可以被分成60000計數(shù)。由于2?5ms的延遲和5％的誤差，這時適當(dāng)?shù)脑O(shè)置精度為1.5°。通過計算，當(dāng)角度增加1.5°PWM值增加40。所以可以如下得出：</p><p>　　PWMDTY01= 4500+40 index_angle</p><p>　　Index_angle=I=index ccd-index_center和

17、index_center是17.indel_ccd代表在黑線位置。如果INDEX_angle值大于0，則當(dāng)前黑線是在智能汽車的中心線右側(cè)的。PWMDTY01的值可以通過上述等式來計算，所述信號驅(qū)動器轉(zhuǎn)向齒輪到右側(cè)。反之，如果index_angle值小于0，PWMDTY01的值驅(qū)動所述轉(zhuǎn)向器轉(zhuǎn)向左側(cè)。</p><p>　　在運行過程中，智能小車花費大量時間在跟蹤直線。通過實驗，我們可以發(fā)現(xiàn)，智能車將劇烈擺動，影響智

18、能車的速度。如果智能車跑的路程直線過長，這將是糟糕的時候。因此，調(diào)整該系統(tǒng)轉(zhuǎn)彎策略。當(dāng)index_ccd值接近中心線，轉(zhuǎn)動角是在小范圍內(nèi)限定。如果角度在7-25指涉點之間，則在大的角度內(nèi)調(diào)整。調(diào)整后的曲線如圖5所示。</p><p><b>　　C.自行控制</b></p><p>　　導(dǎo)向線由四種不同類型的曲線，包括線性，90度角的曲線，大S曲線，和小S曲線。為了達

19、到最快的速度，該系統(tǒng)已經(jīng)開發(fā)了以下的控制策略。</p><p>　　智能小車的自我驅(qū)動策略是基于二維的3*35陣列上。這意味著該系統(tǒng)能檢測已捕獲的黑線做出快速行動來調(diào)整智能的轉(zhuǎn)動角度和速度。</p><p>　　通過大量的實驗，當(dāng)智能汽車運行在大的S曲線時候，可以發(fā)現(xiàn)該行的參考點在行和行之間移動[1] [27]。與此同時，因為高速行駛速度，大S的小弧度，和它的前瞻性能力小車可以以近似直線的

20、方式在大S曲線上運行。當(dāng)智能車進入小S曲線或90度的直角彎，參考點線或行[1] [35]將確定的黑線，這意味著黑色引導(dǎo)線出現(xiàn)在邊緣最遠(yuǎn)的線路。</p><p>　　如果小車?yán)^續(xù)在小轉(zhuǎn)角和高速對應(yīng)的狀態(tài)下運行，智能車將沖出賽道或大幅擺動。在這一點上，系統(tǒng)的策略是交出控制線[2]，它是說基于數(shù)據(jù)的中間一行自我驅(qū)動控制的速度和轉(zhuǎn)向角度。通過擬合曲線，我們可以看到該行[2]的參考點的轉(zhuǎn)角更尖銳，而且速度比前行[1]低。智

21、能車可以迅速改變其塑像，以保持它運行到軌道。如果行的黑點[2] [1]或線路的黑線[2] [35]設(shè)置為1，則進一步的策略是交出控制線的參考點[3]，以降低車速，增加的角度范圍。</p><p>　　驅(qū)動通過曲線后智能車將行駛到軌道上的直線。該如何從曲線控制改變策略到直線控制就顯得尤為重要。為此，該系統(tǒng)引入了計時器的概念：當(dāng)智能汽車的行駛距離是約5cm，速度為3m/ s的范圍內(nèi)（這意味著在0.015s），線[1]

22、 [17]將連續(xù)檢測，直到找到黑線。該系統(tǒng)將把它作為智能車返回到比賽筆直的道路處理，并給回線的控制權(quán)[1]。</p><p><b>　　速度控制和彎道控制</b></p><p>　　它是用于智能汽車，以增加貼近控制的實際值，包括轉(zhuǎn)向角的多少和行駛速度，來在智能汽車駕駛中更接近所需速度的重要因素。因此，采用開環(huán)控制理論和閉環(huán)控制理論控制汽車。</p>

23、<p><b>　　開環(huán)控制系統(tǒng)[6]</b></p><p>　　在開環(huán)控制系統(tǒng)裝置中，控制對象的輸出（控制量），不影響在此控制系統(tǒng)中的控制器輸出。它不依賴于反饋形成任何封閉環(huán)。</p><p><b>　　閉環(huán)控制系統(tǒng)[7]</b></p><p>　　閉環(huán)控制系統(tǒng)的特征在于，該系統(tǒng)控制對象的輸出（控制量）反

24、饋影響控制器的輸出。</p><p><b>　　A.旋轉(zhuǎn)控制</b></p><p>　　當(dāng)智能車行駛過程中遇到需要使用的控制轉(zhuǎn)向策略后，該MC9S12DG128可以通過PWM信號實現(xiàn)對轉(zhuǎn)向的控制。該系統(tǒng)采用開環(huán)伺服控制來實現(xiàn)平穩(wěn)，快速的轉(zhuǎn)向控制。</p><p>　　盡管在實現(xiàn)所需的角度值時，閉環(huán)控制PID算法優(yōu)于開環(huán)控制，但選擇不當(dāng)PID

25、參數(shù)可以導(dǎo)致容易過沖，使得智能車搖擺急，齒輪不順利工作。同時，在用于智能汽車的跟蹤的時候，轉(zhuǎn)向齒輪的旋轉(zhuǎn)總是從一個方向到另一個方向。旋轉(zhuǎn)也不會出現(xiàn)突然逆轉(zhuǎn)的情況。開環(huán)控制的是從偏離所希望的值接近期望值，這功能是適合于轉(zhuǎn)向控制的。因此，系統(tǒng)選擇開環(huán)控制來實現(xiàn)轉(zhuǎn)向控制。</p><p><b>　　B.速度控制</b></p><p>　　智能汽車使用步進電機作為動力裝置

26、。在MAX33886系統(tǒng)中可以通過輸入各種占空比的電信號控制電機的轉(zhuǎn)速。為了使智能車輛迅速達到所需的值，在最短的時間內(nèi)實現(xiàn)增長，減速，該系統(tǒng)采用閉環(huán)控制模式，以調(diào)整速度。</p><p>　　C.增量式PID控制</p><p>　　在實踐中，PID調(diào)節(jié)被廣泛用于閉環(huán)控制[8,9]，這是比例，積分，微分控制。隨著控制理論的完善，有一個增量式PID控制。增量式PID控制算法</p>

27、;<p>　　推導(dǎo)如下：由遞推公式</p><p><b>　?。?）</b></p><p>　　它們時與采樣周期，比例系數(shù)，積分時間常數(shù)，微分時間常數(shù)有關(guān)的系數(shù)。如可以看到的，一般的計算機控制系統(tǒng)采用恒定的采樣周期T，當(dāng)KP，KI，Kd值設(shè)定，只要得到的作為使用偏差的前三次測量參數(shù)就可以。</p><p>　　在增量式PID處

28、理的過程中，有一個步驟在你得到U（K）后。你輸入的PWM到電機時，則必須判斷U（k）的值。如果該值小于0，則輸入PWM信號為0，如果它大于最大PWM信號，則輸入的最大值。</p><p>　　該系統(tǒng)使用一個增量PID算法，公式如下</p><p>　　error = speed_v- infrared_value 7</p><p>　　pwmtemp = PWMD

29、TY 23 + PID_P*（error-last_error）+PID_I*（error）+PID_D（error+pre_error-2*last_error）</p><p>　　在該式中，speed_v代表標(biāo)準(zhǔn)速度，infrared_value代表實時從ATD1轉(zhuǎn)換速度值。計算它們的誤差的差異，并使用增量式PID控制算法計算得到pwmtemp。</p><p>　　該pwmtemp

30、是作為輸入信號來驅(qū)動電動機。因此，PID算法的主要功能是通過實時反饋速度，以使速度接近所希望的速度盡可能的閉環(huán)系統(tǒng)。</p><p>　　就是說，該系統(tǒng)可以從路徑所需的速度確定模塊和實時速度形成速度檢測模塊，然后調(diào)整PWM信號，以適應(yīng)不同的路徑的條件。 </p><p><b>　　D.魯棒控制策略</b></p><p>　　通過使用增量PI

31、D調(diào)節(jié)，我們可以迅速調(diào)整模型車輛的速度到理想值。然而，實驗表明，當(dāng)模型車突然拐彎，從一條直線或運行到一個繞來繞去彎道后，不能加快速度或立即減速。這意味著PID調(diào)節(jié)在這種情況下不是特別敏感。鑒于此，本文提出了一種被稱為魯棒控制策略，用他代替增量PID調(diào)節(jié)法。同時魯邦可以實現(xiàn)速度調(diào)節(jié)的輔助策略。</p><p>　　魯棒控制裝置是一種快速優(yōu)化控制。簡單地說，它提供了最大馬力和最大的制動力。例如，從A點行駛時到B點，以

32、最快的方式是，我們應(yīng)該給車子的最高速度，不考慮轉(zhuǎn)彎率，障礙等因素有關(guān)。當(dāng)汽車到達終端，直接的方式是停止提供最大的制動力。</p><p>　　在本文中，魯棒控制策略的使用方法如下：進入直路后，系統(tǒng)分配的PWM占空比來最大，也就是電機運行的最高速度。速度盡快達到所需的值作為小車運行到直線的時候。當(dāng)速度達到閾值，則系統(tǒng)改變控制策略，以PID控制平穩(wěn)地調(diào)整速度。當(dāng)智能汽車轉(zhuǎn)彎到一角落，該系統(tǒng)減少了PWM占空比為零，即切

33、削電機的功率，從而使智能車速度能迅速降低。當(dāng)速度達到所述閾值的PID控制被啟用。</p><p>　　智能車開始可以用最快的速度在直道上，當(dāng)汽車轉(zhuǎn)彎的拐角處，趕緊減慢速度。與此同時，對于其他情況，智能車在魯棒控制策略輔助下平穩(wěn)運行。由于魯棒控制策略，該系統(tǒng)提高了汽車的速度，并避免了“運行停止”。</p><p><b>　　五，實驗結(jié)果</b></p>

34、<p>　　系統(tǒng)測試是在封閉的實驗室環(huán)境中進行的，我們使用的是大學(xué)智能賽車競賽“飛思卡爾杯”的指定路徑。使用的光源是普通熒光燈。測試包括三個方面，如下：1）</p><p>　　如下:1)測試CCD傳感器，通過上位機的觀察數(shù)據(jù)來確認(rèn)其是否可以準(zhǔn)確的收集路徑信息。2）測試閉環(huán)控制它是否可以成為調(diào)整速度快于開環(huán)控制電動機的方法。3）測試當(dāng)智能汽車運行在跟蹤路徑上時，齒輪的轉(zhuǎn)向是否靈敏和順暢。</p&g

35、t;<p>　　如圖7所示,通過有藍牙模塊的串口工具，我們可以清楚的在主機中看到路徑信息[10]。由此可見,智能汽車可以準(zhǔn)確的收集并提取路徑信息。</p><p>　　從圖8和圖9中，不難發(fā)現(xiàn)，與開環(huán)控制相比，使用PID比閉環(huán)控制可以使汽車達到所需的值更快。這兩張照片都顯示過程如何智能車達到所需的值。</p><p>　　該智能車被放置在跑道上運行十圈。可以發(fā)現(xiàn)，該汽車可以非

36、常好的適應(yīng)各種路徑，并且行駛路徑平滑。它也出現(xiàn)了“RUN-STOP”現(xiàn)象。智能車速度是通過使用電子計時裝置和其它輔助設(shè)備計算的，結(jié)果是，平均速度可達3米/秒。</p><p><b>　　六．結(jié)論</b></p><p>　　本文首先簡要介紹了系統(tǒng)的總體設(shè)計目標(biāo)和結(jié)構(gòu)。然后根據(jù)對整個系統(tǒng)的需求，系統(tǒng)地分為了不同的功能模塊。每個模塊的設(shè)計實施以及智能控制思想都在該<

37、;/p><p>　　模塊進行了詳細(xì)的描述。最后，進行了最終的系統(tǒng)使用功能試驗。通過試驗結(jié)果，</p><p>　　我們可以看到，智能小車具有高度自適應(yīng)能力和自我跟蹤。結(jié)果表明，智能控制</p><p>　　基于智能車的研究是成功的。</p><p>　　結(jié)合了智能控制技術(shù)和嵌入式技術(shù)的設(shè)計將會有更廣闊的應(yīng)用前景，不僅如此，它還將促進社會的進步并且

38、帶來巨大的經(jīng)濟效益。</p><p>　　附件2：外文原文（復(fù)印件）</p><p>　　Intelligent Control Research based on the Smart Car</p><p>　　Abstract: This paper uses a “smart car”as the platform for intelligent contro

39、l research. MC9S12DG128 is chosen to be the main control unit, which is produced by Freescale Semiconductor. A smallest intelligent control system is designed in this paper and it implements smart car’s self-tracking dri

40、ving using path identification algorithm. In this paper, the intelligent control study of the smart car includes: the extraction of path information, self-tracking algorithm implementation and direction </p><p

41、>　　Key words: MC9S12DG128； Intelligent Control ；Open-loop control ；PID ；BangBang；</p><p>　　I. INTRODUCTION</p><p>　　With the improvement of control theory, as well as the rapid development of

42、 information technology, intelligent control plays an increasingly important role in our society.Embedded devices have small sizes, low power consumption, powerful functions and ect. There will be arelatively wide range

43、of applications in this field, such as automotive electronics, aerospace, smart home and it will spread to other fields if these technologies work together. In order to study the embedded intelligent contr</p><

44、;p>　　At first, this paper gives readers a general introduction of the smart vehicle system [2, 3]. Then, according to the process of the smart vehicle intelligent control including: the extraction of path information,

45、 self-tracking algorithm implementation, steering engine direction and speed control,it gives the detail information of the above four aspects of the smart car system. In addition, the paper highlights the control proces

46、s of the smart car application including open-loop control, closed-l</p><p>　　II. GENERAL DESIGN OF SMART CAR SYSTEM</p><p>　　The system uses MC9S12DG128 [4] as the master chip, </p><

47、p>　　and a CCD sensor as the traffic information collecting </p><p>　　sensors. The speed sensor is developed based on the principle of radio-type photoelectric tube. The path can be drawn after CCD sensor

48、collected the data and the system figures out the corresponding process. At the same time, the smart car's current speed which is measured by the electric motor for speed test module is responded to the system.Finall

49、y, the path identification system utilizes the path information and the current speed to make the smart car run at the highest speed in different roa</p><p>　　III. THE IMPLEMENTATION OF SELF-TRACKING ALGORIT

50、HM</p><p>　　The smartcar’s self-control is baseed on the path </p><p>　　information which is collected by the CCD sensor [5]. CCD</p><p>　　sensor’s data-acquisition rate is 25frames

51、/s. One frame is divided into two parts: Odd Fields and Even Fields, that is to say, 50fields/s. In order to make an accurate judgment of the current path we amplify the frequency to 32MHz. Finally, the system will get a

52、 35*307 array after MC9S12DG128 ATD module converse the data which are collected by the CCD sensor and ignore some blanking data. </p><p>　　The smart car has hysteretic quality while running. In order to inc

53、rease the speed, collecting forward-looking path information becomes particularly important. For the above reasons, we select the 1st row of the array which is the emote data as basic data for the smart car control.Corr

54、espondingly, the 150th line and 300th line are selected as auxiliary data, and then we get a new two-dimensional array with a 3*35 style. On the basis of the array, we can fit the optimum speed value and the turni</p&

55、gt;<p>　　Speed Curve Fitting</p><p>　　Figure 2 shows the position relationship between the</p><p>　　smart car and the path (black track, exactly) when the car turns around. L denotes the

56、distance between the farthest reference line which the CCD sensor collects and the car. S denotes the distance between intermediate point of the farthest reference line and the track line. R denotes the turning radius.&l

57、t;/p><p>　　The process of calculating R is as follows: </p><p>　　Deriving from trigonometric functions, we work out：</p><p>　　After measurement, l=0.6m。In the 90-degree angle </p>

58、;<p>　　corner, the distance between the black track location and the vertical centerline of the car location is about 0.2667m. It means that S ranges from 0 to 0.2667m.</p><p>　　According to the centr

59、ipetal force formula:</p><p>　　The limit speed can be derived from the formula as </p><p><b>　　followed:</b></p><p><b>　　（4）</b></p><p>　　The sp

60、eed of each reference point which is the maximum value when the smart car enters into the corner can be calculated by the formula. To calculate the speed v, we need to measure the friction parameter.</p><p>

61、;　　In this paper, the method which is used to calculate the </p><p>　　friction parameter is as below: At first, the smart car is placed on the KT board, and then one side of the board is raised until the sm

62、art car can slip from the board. </p><p>　　Measure the angle which is between the ground and the KT</p><p>　　board. The result is about 60° The Figure 3 shows the </p><p>　　mec

63、hanical analysis of the smart car. The following equation can get the physical condition of force balance;</p><p>　　mgsin 60 = mgsin 30 (6)</p><p>　　The friction coefficient of KT board can

64、be calculated,and then μ=0.577</p><p>　　Each row of the image array which the smart car identifies is composed of 35 points. From left to right, the reference point is numbered from 1 to 35. In the paper, t

65、he 0 to 17 points are chosen as examples to declare. The maximum speed of these points is calculated, and these data are showed in the table 1</p><p>　　As can be seen from the table, the smart car’s maximum

66、speed is 2.1596m/s. After testing, the result shows that speed is 2.0m/s when PWM is equal to 2400. When the smart car runs in the straight line before it enters into the corner, the speed may be faster than the maximum

67、 speed. At the same time, the inertia is anther factor for the traveling smart car. The system is tested through many experiments which are based on the maximum value, and finally, the speed curve of the reference point

68、s is </p><p>　　B. Trunning Curve Fitting</p><p>　　Smart car direction is controlled by the steering gear.The range of the steering gear’s turning angel is from -45° to +45°.In this pap

69、er, MCU bus frequency is multiplied to32MHz through the PLL, and 20ms needs 60000 counts.Because of the 2~5ms delay and 5% error, it’s appropriate to set accuracy to 1.5°. By calculating, the PWM value increases 40

70、when the angle increases 1.5°. So the responding equitation can be drawn as below:</p><p>　　PWMDTY 01 = 4500 + 40 index_angle </p><p>　　Index_angle=index _ccd-index_center, and in

71、dex_center is 17.indel_ccd stands for black-line position in the arry. If the index_ angle value is larger than 0, the current black line is on the right side of the smart car’s centerline. The value of PWMDTY01 can be c

72、alculated through the above equation, the signal drives steering gear to the right. On the contrary side, if the index_angle value is less than 0, the value of PWMDTY01 drives the steering gear turn to the left side.<

73、/p><p>　　Smart car spends much time on tracking the straight line in the running process. Through the experiment, we can find that the smart car will tempestuously swing which affects smart car’s speed. It will

74、 be worse when the smart car may run away if the straight line is too long. Therefore, the system adjusts the turning strategy. When the index_ccd value is near the centerline, the turning corner is limited in a small ra

75、nge. If the angle is between the 7-25 referent points,turn in a large angle. Ad</p><p>　　C. Self-driving control</p><p>　　The guide line is composed of four different types of curve, including l

76、inear, 90-degree angle curve, big S curve, and small S curve. In order to achieve the fastest speed, the system has developed the following control strategies.</p><p>　　Smart car’s self-drive strategies are

77、based on a </p><p>　　two-dimensional array of 3*35. It means that the system detect which spot has captured the black-line and make a quick action to adjust the smart’s turning angle and speed. </p>&

78、lt;p>　　Through a large number of experiments, it can be found </p><p>　　that the reference spots of line[1] move between line[1][5]and line [1][27] when smart car runs on large S curve. At the same time,

79、the smart car can run through the large S curve by approximated straight line way because of the high-speed driving speed, the large S’s smaller corner, and its forward-looking ability. When the smart car enters small S

80、curve or 90-degree right-angle corner, the reference point line[1][1] or line[1][35] would identify the black line, which means the black guide line</p><p>　　the further strategy is to hand over control to t

81、he reference spots of line[3] to reduce speed and increase the angle range.</p><p>　　Smart cars will be traveling to the track on a straight line after driving through the curve. It is particularly important

82、 that how to change strategy from curve control to a straight line control. To this end, the system introduces the concept of timer: when the smart car’s driving distance is about 5cm and the speed is within 3m/s( which

83、means in 0.015s), line[1][17] will continuously detect until finding the black line. The system will handle it as the smart car returns to straight race road</p><p>　　IV. SPEED CONTROL AND TRUNNING CONGTOL&

84、lt;/p><p>　　It is an important factor for the smart car to increase the speed that is how to control the actual value including turning angle and driving speed to approach the desired value during the smart car

85、 driving. Therefore, the open-loop control theory and closed-loop control theory was adopted to control the car.</p><p>　　Open-loop control system[6]</p><p>　　Open-loop control system means the

86、controlled object's output (controlled quantity) do not affect the controller output in this control system. It does not rely on the feedback forming any closed-loop. </p><p>　　Closed-loop control system

87、[7]</p><p>　　Closed-loop control system is characterized that the system controlled object’s output (controlled quantity) is sent back to affect the anti-controller's output.</p><p>　　A. Tur

88、ning control</p><p>　　The MC9S12DG128 achieve the steering control through PWM signals after the system had a desired control steering strategy during the smart car running. The system adopts the open-loop s

89、ervo control to implement a smooth, fast steering control. </p><p>　　Although the closed-loop control PID algorithm outperforms the open-loop control in achieving the desired angle value, improper selection

90、of PID parameters can lead to overshoot easily, which makes the smart car swing sharply and steering not gear turn smoothly. At the same time, for the smart cars tracking, the steering gear rotation is always from one di

91、rection to the other direction. Rotation will not appear suddenly reverse situation. The feature of open-loop control which is approaching the</p><p>　　B. Speed control</p><p>　　Smart cars use s

92、tepper motor as a power plant. The </p><p>　　system using MAX33886 can control the motor speed by inputting various duty cycle electrical signals. In order to make the smart vehicles quickly achieve the desi

93、red value,which is the shortest time possible to achieve growth, deceleration, the system uses closed-loop control mode to adjust the speed.</p><p>　　C. Incremental PID control</p><p>　　In pract

94、ice, PID regulation is widely used for closed loop control[8,9], which is proportional, integral, differential control. With the improvement of control theory, there is anincremental PID control. Incremental PID control

95、algorithm is derived as follows: </p><p>　　By the recursive formula</p><p><b>　?。?）</b></p><p>　　They are related to the sampling period, the proportion coefficient, int

96、egral time constant, derivative time constant related to the coefficient. As can be seen, the general computer control system uses a constant sampling period T, when kp, ki, Kd are set, you can get the parameter as long

97、as the use of the three previous measurements of the deviation,They are related to the sampling period, the proportion coefficient, integral time constant, derivative time constant related to the coefficient.</p>

98、<p>　　the use of the three previous measurements of the deviation。</p><p>　　In the process of incremental PID treatment, there is a step to note after you get u(k). Before you input the PWM to the moto

99、r, you must judge the u(k ) value. If the value is less than 0, input the PWM signal as 0, if it is greater than the maximum PWM signal, input the max value.</p><p>　　The system uses an incremental PID algo

100、rithm, the </p><p>　　formula is as follows：</p><p>　　error = speed_v- infrared_value 7</p><p>　　pwmtemp = PWMDTY 23 + PID_P*（error-last_error）+PID_I*（error）+PID_D（error+pre_error-2*

101、last_error）</p><p>　　In the formula, speed_v stands for the standard speed, infrared_value7 means speed value of real-time from ATD1 </p><p>　　conversion. The pwmtemp is assigned after calculati

102、ng the </p><p>　　difference of their error and using incremental PID control algorithm to calculate pwmtemp.The pwmtemp is as an input signal to drive the motor.So, the main feature of PID algorithm is using

103、 real-time feedback speed to make speed approaching the desired speed as far as possible in the closed-loop system.</p><p>　　It is said that the system can get the desired speed </p><p>　　from t