智能車外文翻譯---自治智能車在模擬車輛列隊中的設(shè)計

1、<p>　　附件1：外文資料翻譯譯文</p><p>　　自治智能車在模擬車輛列隊中的設(shè)計</p><p>　　萬建，楚秀敏，吳勇，張瑞</p><p>　　運輸安全工程研究中心，教育部，</p><p>　　武漢理工大學(xué)，武漢，湖北，430063，中國。</p><p>　　E-mail:whut_wj@1

2、26.com</p><p><b>　　摘要</b></p><p>　　自治智能車是基于考慮車輛和道路在內(nèi)的車輛編隊的物理仿真的基礎(chǔ)。本文在車輛道路綜合的情況下，分析了車輛編隊系統(tǒng)的架構(gòu)，并提出了自制智能車控制系統(tǒng)的構(gòu)造和結(jié)構(gòu)。在分析了自治智能車的功能要求之后，本文設(shè)計了自治智能車關(guān)鍵的硬件和軟件。它把芯片作為控制器，以及用攝像頭和超聲波傳感器作為行車導(dǎo)航。同時，

3、它應(yīng)用直流電機實現(xiàn)智能車的驅(qū)動和轉(zhuǎn)向，以及采用Zigbee技術(shù)來設(shè)計無線通信模塊。我們提出的關(guān)于識別導(dǎo)航線和運動控制的關(guān)鍵算法，這其中包括路徑提取和控制算法。試驗表明自治智能車有一個良好的穩(wěn)定性能，滿足了車輛編隊系統(tǒng)的功能要求，這款車將提供測試平臺和車輛編隊系統(tǒng)的進一步研究的技術(shù)基礎(chǔ)。</p><p><b>　　1.簡介</b></p><p>　　近年來，隨著橫向

4、和縱向的智能車輛控制技術(shù)等智能交通技術(shù)的發(fā)展，車輛列隊研究已成為在智能交通領(lǐng)域的熱點，它融合了一些技術(shù)，這其中包括車輛間相互通信，公路通信技術(shù)，智能控制技術(shù)等。在車輛道路綜合的基礎(chǔ)上的車輛隊列控制系統(tǒng)可以通過提高單個車輛的智能化水平，提高與交通環(huán)境交互信息的能力，以及增加車輛密度來提高道路通行能力。與此同時，它減少了控制對象，簡化了交通控制復(fù)雜性，增加了運輸可控性，有效地緩解了交通堵塞，并最終提高了行車安全性。此外，它可以在一定程度上減

5、少車輛阻力和車輛油耗。</p><p>　　圖1顯示了基于車輛道路行駛的車輛隊列架構(gòu)系統(tǒng)，這表明智能車輛控制，車路信息交互技術(shù)，車輛隊列和控制方式和其他關(guān)鍵技術(shù)是系統(tǒng)的重要組成部分。然而，目前汽車隊列的結(jié)構(gòu)，行為特征和智能化行程控制算法尚未完善。因此，有必要研究一些基礎(chǔ)東西，這包括車輛隊列，車輛隊列模型，及車輛小隊控制方法的行為特征，這些研究需要在建設(shè)有硬件循環(huán)仿真的車輛隊列系統(tǒng)中進行。</p>&

6、lt;p>　　圖1基于車輛道路的車輛隊列系統(tǒng)的體系結(jié)構(gòu)</p><p>　　為實現(xiàn)對隊列行駛車輛的模擬,智能控制和交換信息是必要的。該系統(tǒng)需要智能控制,信息交互、自治的穩(wěn)定性。圖2是的自治智能車系統(tǒng)的結(jié)構(gòu)。自治智能車采用單片機作為控制器，并使用攝像頭、導(dǎo)航傳感器和超聲波傳感器，基于Zigbee技術(shù)的無線通信模塊。</p><p>　　本文將首先分析基于自治智能車功能要求的整體設(shè)計方案

7、，然后介紹了橫向和縱向的控制和導(dǎo)航的硬件實現(xiàn)方式和如何處理的關(guān)鍵問題，并討論如何通過優(yōu)化控制算法和軟件來提高汽車的穩(wěn)定性和智能化水平，隨后根據(jù)實際過程的測量自治智能汽車的性能介紹它的自治的策略。</p><p>　　圖2自治智能小車控制系統(tǒng)</p><p>　　2. 自治智能模擬車整體設(shè)計</p><p>　　自治智能車包括四個部分：檢測系統(tǒng)，電力系統(tǒng)，通信系統(tǒng)，控

8、制和決策系統(tǒng)。</p><p>　　檢測系統(tǒng)是最重要的，其主要工作是導(dǎo)航。我們選擇CMOS攝像機作為檢測傳感器，它可以檢測出車道，引導(dǎo)車輛在路上順利的行駛?？紤]到單一的CMOS圖像傳感器不能理想的檢測車的距離，超車距離和其他運動參數(shù)，我們選擇了超聲波傳感器。</p><p>　　電力系統(tǒng)主要控制電機的轉(zhuǎn)向角度和直流電機的速度。該控制過程如下：首先，檢測系統(tǒng)收集路徑信息，然后驅(qū)動器系統(tǒng)使直流

9、電動機產(chǎn)生適當(dāng)?shù)霓D(zhuǎn)速，轉(zhuǎn)向電機根據(jù)控制和決策系統(tǒng)的分析和判斷給出一個正確的轉(zhuǎn)向角，所以自治智能汽車可以跑得快而且平穩(wěn)。為了通過車隊仿真控制得到更實際的行為特征,自治智能車在該系統(tǒng)中采用后輪驅(qū)動和前輪轉(zhuǎn)向的結(jié)構(gòu)。</p><p>　　在車輛與車輛之間和車輛與道路之間的無線通信網(wǎng)絡(luò)中，我們選擇了基于IEEE802.15.4無線標(biāo)準(zhǔn)的ZigBee技術(shù)。多節(jié)點的網(wǎng)絡(luò)需要大的網(wǎng)絡(luò)容量和自組織無線通信技術(shù)，相較于其他無線通訊

10、技術(shù),Zigbee有更高的數(shù)據(jù)傳輸速率和執(zhí)行時的更穩(wěn)定。它的網(wǎng)絡(luò)能夠支持上千個節(jié)點，是在小范圍內(nèi)監(jiān)測和控制的最好選擇。</p><p>　　控制和決策系統(tǒng)以飛思卡爾16位單片機- MC9S12XDP512作為其主要控制芯片，它的40M的主頻能夠滿足實時檢測和信息處理的需要。此外，它可以提供豐富的I/ O端口，精確的時鐘輸入捕捉和時鐘資源。XDP512連接所有的子模塊，收集周圍的車輛信息，并處理數(shù)據(jù)，根據(jù)控制算法輸

11、出控制信號。</p><p><b>　　圖3系統(tǒng)框架</b></p><p>　　3.自治智能車硬件系統(tǒng)設(shè)計</p><p>　　硬件設(shè)計對自治智能車的運行效果有直接的影響，根據(jù)自治智能車總體的分析，硬件應(yīng)該含??有以下模塊：（1）導(dǎo)航模塊，這其中包括數(shù)字CMOS攝像頭和超聲波無障礙檢測傳感器（2）控制模塊，包括行駛和轉(zhuǎn)向運動電機（3）Zig

12、bee無線通信模塊（4）單片機控制模塊。</p><p><b>　　3.1、導(dǎo)航裝置</b></p><p>　?。?）攝像頭模塊設(shè)計</p><p>　　使用攝像頭作為路徑檢測傳感器提前掃描前面的道路,以便汽車更順暢的操作。由于CMOS圖像傳感器具有高集成度，低功耗，低像素的缺陷和其他優(yōu)勢，我們選擇了356* 292分辨率OmniVisio

13、n的OV6620彩色CMOS圖像傳感器。圖4是圖像采集過程。首先,單片機控制攝像頭采集信息，然后傳輸圖像數(shù)據(jù)到FIFO緩沖存儲器，變換并行及串行數(shù)據(jù),最后由單片機的SPI端口讀取這個數(shù)據(jù)。</p><p><b>　　圖4 圖像采集過程</b></p><p>　　圖像采集過程有兩種模式：上電模式和SCCB模式。該系統(tǒng)采用SCCB模式：經(jīng)過SCCB初始化OV6620和

14、啟用VSYNC，系統(tǒng)判斷是否已獲得一幀圖像，F(xiàn)IFO存儲了一幀圖像之后，系統(tǒng)通過單片機獲取數(shù)據(jù)。</p><p>　?。?）超聲波模塊設(shè)計</p><p>　　如果讓智能汽車能自動避開障礙和導(dǎo)航，則它需要建立在行駛中車輛的距離測量系統(tǒng)。超聲波測距系統(tǒng)，可避開障礙并定位他們，根據(jù)攝像頭獲得的信息進行決策控制，并協(xié)助路徑的規(guī)劃。但是少量的超聲波傳感器不能滿足高精度測距和避障的要求。為了確保高精

15、確度，汽車需要增加測量距離的電路通道，用來補償傳感器角度的限制。該系統(tǒng)采用8個超聲波測量通道，利用角度補償手段使主要障礙的位置和距離的信息更加準(zhǔn)確。圖5是安裝結(jié)構(gòu)圖，圖6是距離測量系統(tǒng)的數(shù)據(jù)流結(jié)構(gòu)圖。</p><p>　　圖5超聲波傳感器安裝結(jié)構(gòu)</p><p>　　該系統(tǒng)通過檢測從發(fā)射到返回的時間間隔來計算距離。因為時間與超聲波的路程成正比，當(dāng)超聲波發(fā)射端發(fā)送幾個振蕩的脈沖，微控制器開始

16、計時；當(dāng)接收器接收到第一個反饋脈沖，時間停止。測量距離如下：</p><p>　　D=CT/2。（1）</p><p>　　在公式（1），C是空氣中的聲速，T為從發(fā)射到返回的時間間隔。圖6如下：</p><p><b>　　3.2 控制單元</b></p><p>　　（1）直流電機驅(qū)動，速度檢測</p>

17、<p>　　RS -380S型直流電動機是用于速度控制，自治智能車采用閉環(huán)控制技術(shù)，并以MC33886 H橋驅(qū)動器作為電機驅(qū)動器。如果電機采用開環(huán)控制，它會受到許多干擾，如電池電壓，電氣傳動摩擦，路面摩擦力和由前輪轉(zhuǎn)向角引起正向電阻，這些因素將導(dǎo)致智能車的運行不穩(wěn)定。因此，閉環(huán)控制方法是迫切需要的。閉環(huán)控制系統(tǒng)測量速度，并采用PID算法，它需要在很短的間隔內(nèi)獲得速度變化，計算出瞬時速度和期望速度之間的差值，速度傳感器采用歐姆龍

18、E6A2- CWZ3C編碼，其精度可達(dá)360 P/ R。</p><p><b>　?。?）轉(zhuǎn)向電機控制</b></p><p>　　轉(zhuǎn)向電機控制由直接改變輸入PWM占空比的不同來轉(zhuǎn)動不同的角度，該轉(zhuǎn)向電機輸出角與給定的PWM信號有一定的線性關(guān)系。由于電機的轉(zhuǎn)向力矩足夠大，單片機計算橫向控制量，并直接給出了PWM控制信號，使電機實現(xiàn)轉(zhuǎn)向。</p><

19、;p><b>　　3.3無線通信模塊</b></p><p>　　自治智能車的通信系統(tǒng)包括以ZigBee為基礎(chǔ)的通信衛(wèi)星網(wǎng)絡(luò)，它由一個網(wǎng)絡(luò)協(xié)調(diào)器和一些網(wǎng)絡(luò)終端節(jié)點組成。網(wǎng)絡(luò)協(xié)調(diào)員負(fù)責(zé)網(wǎng)絡(luò)的管理工作，而終端節(jié)點一方面獲得模擬數(shù)據(jù);在另一方面，把這些模擬數(shù)據(jù)通過無線網(wǎng)絡(luò)傳輸給協(xié)調(diào)員。通過這種方式，不僅降低了ZigBee網(wǎng)絡(luò)的復(fù)雜性，而且也方便了數(shù)據(jù)的集中管理。圖7描述了通信網(wǎng)絡(luò)的設(shè)計方案效

20、果：在PAN無線網(wǎng)絡(luò)覆蓋里，網(wǎng)絡(luò)終端節(jié)點的數(shù)目已經(jīng)確定。</p><p>　　在這個系統(tǒng)中，每臺車是一個智能終端節(jié)點。裝載著MC13192無線收發(fā)器的汽車通過MC13192與 XDP512之間的數(shù)據(jù)交換來進行無線通信。</p><p>　　圖7 自治智能車通信系統(tǒng)結(jié)構(gòu)</p><p>　　3.4．單片機及其外圍電路模塊</p><p>　　X

21、DP51是自治智能車的核心部件,它控制著所有其他模塊,獲得路徑、速度、無線信號和其他數(shù)據(jù),并在此基礎(chǔ)上將參數(shù)歸類并計算出最優(yōu)的控制策略。因此,該系統(tǒng)必須擁有非常高的穩(wěn)定性。</p><p>　　為了提高單片機的穩(wěn)定性，設(shè)計的主要措施已經(jīng)采取如下：①MCU電源電路設(shè)計;②濾波電路的優(yōu)化;③單片機系統(tǒng)PCB板的布局;④單片機時鐘電路。</p><p>　　4．自治智能車和相關(guān)算法的軟件系統(tǒng)設(shè)計

22、</p><p>　　該智能車系統(tǒng)軟件包括以下模塊：初始化模塊，實時路徑檢測模塊，防沖突模塊，橫向和縱向閉環(huán)控制模塊，通信模塊。該系統(tǒng)的軟件流程圖如圖8所示。在大多數(shù)時候，微控制器處理數(shù)據(jù)和圖像。因此，高效的圖像處理算法和閉環(huán)控制算法可以節(jié)省單片機的CPU資源，提高自治智能車的反應(yīng)速度和它的整體性能。路徑提取算法和運動圖像采集后的反饋控制算法如下。</p><p><b>　　4

23、.1路徑提取算法</b></p><p>　　路徑是目標(biāo)檢測線邊緣的準(zhǔn)則。算法是：一個灰度圖像中的每個設(shè)定的閾值的二維矩陣，得到兩個相鄰像素的自頂向下的差值。如果邊緣大于或等于閾值，它的下一個點對應(yīng)的像素是指向邊緣的，該像素被認(rèn)為是特征點，在同一時間記錄它。當(dāng)發(fā)現(xiàn)邊緣的排列，我們可以找出靠近這一行的下一行近的邊緣，因此花更少的時間找到了這一點。該算法能始終在每一列的邊緣附近跟蹤這一列,并找出下一行的邊

24、緣,所以它是高效的。在橫向控制，我們根據(jù)坐標(biāo)來使電機轉(zhuǎn)向，該查表方法可以控制橫向方向。</p><p>　　圖8 自治智能車系統(tǒng)的軟件流程圖</p><p>　　4.2運動反饋控制算法</p><p>　　在縱向控制中，我們建立一個二維數(shù)組，有10*33種元素，每一行對應(yīng)一個速度值，在一定的速度下，每個排列對應(yīng)著不同的角。</p><p>　

25、　在縱向控制的過程中，我們根據(jù)當(dāng)前的速度和道路狀況設(shè)置安全速度值，所謂的安全速度值是車能拐過拐角的速度。當(dāng)反饋速度小于設(shè)定速度，汽車加快速度，如果反饋速度等于設(shè)定速度，汽車保持原來的狀態(tài)，否則減慢或加快。加速和減速算法公式（2）：</p><p>　　u(k)=Kp[e(k)-e(k-1)]+Kie(k)+Kd[e(k)-2e(k-1)+e(k-2)] (2)

26、 </p><p>　　△u(k)是速度變量的增值、e(t)是控制誤差、Kp相當(dāng)增益,Ki= KpT /Ti是積分系數(shù),</p><p>　　Kd = KpTd/T是微分系數(shù)，Ti是積分時間常數(shù)，Td是導(dǎo)數(shù)時間常數(shù)，T是采樣時間。</p><p>　　5．自治智能車測試和分析</

27、p><p>　　如圖9所示的是智能車的外觀，我們從四個方面測試它的性能表現(xiàn)。</p><p><b>　　圖9智能車的外觀</b></p><p><b>　　5.1智能車的巡線</b></p><p>　　經(jīng)調(diào)整后，智能車的巡線功能實現(xiàn)了。它會自動加速和減速。用計算機模擬的軌跡圖，如圖10，我們可以看

28、到，車開動的路線，有一定的橫向誤差。分析發(fā)現(xiàn)，橫向控制數(shù)據(jù)不夠精細(xì)，那么方向的控制是不能滿足控制精度。</p><p>　　圖10 自治智能車運動軌跡</p><p>　　5.2 汽車蔽障測試</p><p>　　當(dāng)自治智能車需要改變車道或超車，在避開障礙的實驗過程它可以自動避開障礙物.分析表明有關(guān)蔽障策略不能良好的處理速度、距離和轉(zhuǎn)角的關(guān)系。精確模型應(yīng)建立在其中

29、。</p><p>　　5.3單個智能車速度控制的測試</p><p>　　在公示（2）里面改變Kp,Ki,Kd的值，測量電機控制和智能汽車在高速運行的關(guān)系，這種關(guān)系見圖11。橫坐標(biāo)是測量周期和Y坐標(biāo)是測量速度的脈沖，曲線1表示對象速度，曲線2，曲線3是當(dāng)Kp,Ki,Kd變化時的速度曲線。在實驗中，我們發(fā)現(xiàn)系統(tǒng)受Kp影響非常大，圖11顯示，在調(diào)整過程的速度的過程中，由于不同的PID參數(shù)值，

30、會出現(xiàn)不同幅度的振動。特別是在響應(yīng)速度下降時，會產(chǎn)生更大的穩(wěn)態(tài)誤差。</p><p>　　圖11電氣特性和是將時間的關(guān)系</p><p>　　5.4反干擾和通訊測試</p><p>　　在正常情況下，自治智能車在軌道線上運行時，在不脫軌的前提下速度比安全速度小。在增加人為光線或覆蓋一些道路標(biāo)記，自治智能汽車能在遇到盲點時自動停止，所以系統(tǒng)需要通過增加反干擾能力提高性

31、能。同一時間，在自治的智能車上進行的測試表明：通信系統(tǒng)可以正確地接受指令，做出正確的動作。</p><p><b>　　6．結(jié)論與展望</b></p><p>　　基于車輛道路綜合情況的車輛隊列控制是智能交通領(lǐng)域的熱點，半實物仿真技術(shù)是車輛隊列控制的重要研究工具。汽車的通信能力是車輛隊列的物理模擬仿真系統(tǒng)的基礎(chǔ)。在本文中，16位芯片MC9S12X- DP1512是用于

32、控制的核心；除此之外，CMOS攝像頭ov6620傳感器與超聲波傳感器，用于收集交通信息；直流電動機及其它元件組成自治智能車的控制系統(tǒng)；Zigbee技術(shù)是用于通信，這種通信符合單輛車在車隊中的智能化，信息化，自治性，穩(wěn)定性的要求。試驗表明，自治智能車可以自動識別路徑，在高速運行時保持穩(wěn)定性。在通信中，Zigbee數(shù)據(jù)傳輸模塊傳輸數(shù)據(jù)穩(wěn)定、正確，這樣自治智能車可以根據(jù)通信協(xié)議控制另一個智能車。超聲波傳感器有4毫米的位置精度可有效檢測周圍的障

33、礙。汽車的結(jié)構(gòu)提供了用于智能車輛道路系統(tǒng)和實施自治車輛隊列控制的下一步發(fā)展測試平臺和技術(shù)基礎(chǔ)。</p><p>　　今后，將進一步研究自主智能車，車輛隊列控制器算法和控制策略，這項研究包括自治智能車自動跟蹤算法和自動避障算法，多輛車之間的通信，車輛動力學(xué)模型和運動模型的控制策略的結(jié)合。</p><p><b>　　鳴謝</b></p><p>

34、　　它是由國家自然科學(xué)基金項目中國科學(xué)基金（No.50578128）和中國高新高技術(shù)研究發(fā)展計劃（863）（編號2006AA11Z215）。</p><p>　　Tú 2 shì de zìzhì zhìnéng chē xìtǒng de jiégòu. Zìzhì zhìnéng

35、chē cǎiyòng dānpiànjī zuòwéi kòngzhì qì, bìng shǐyòng shèxiàng tóu dǎoháng chuángǎnqì hé chāoshēngbō chuángǎnqì,Zigbee jìsh&#

36、249; de wúxiàn tōngxìn mókuài.</p><p><b>　　字典</b></p><p><b>　　名詞 </b></p><p><b>　　介紹</b></p><p><b>　　

37、采用</b></p><p><b>　　導(dǎo)言</b></p><p><b>　　引言</b></p><p><b>　　導(dǎo)論</b></p><p><b>　　緒論</b></p><p><b>　　

38、引導(dǎo)</b></p><p><b>　　緒言</b></p><p><b>　　初步</b></p><p><b>　　例言</b></p><p><b>　　聿</b></p><p><b>　　諸

39、言</b></p><p><b>　　附件2：外文原文</b></p><p>　　The Design of Autonomous Smart Car Used in Simulation of Vehicle Platoon</p><p>　　Wan Jian, Chu Xiumin, Wu Yong, Zhang Rui

40、 </p><p>　　Engineering Research Center of Transportation Safety, Ministry of Education, </p><p>　　Wuhan University of Technology, Wuhan, Hubei, 430063, China. </p><p>　　E-mail:whut_

41、wj@126.com</p><p><b>　　Abstract</b></p><p>　　The autonomous smart car is the foundation of physical simulation of vehicle platoon based on vehicle and road cooperation. This paper an

42、alyzed the architecture of vehicle platoon system in the case of vehicle-road cooperation, and proposed the constitution and structure of autonomous smart car control system. After analyzing functional requirement of the

43、 autonomous smart car, the paper designed the key hardware and software of the autonomous smart car. It took the microchip as the controller, a</p><p>　　recognizing navigation lane and movement controlling m

44、ethod was proposed, including path extraction and controlling algorithms. The test indicated the autonomous smart car had a good and stable performance, which met functional requirement of vehicle platoon system. The car

45、 will provide test platform and technological base for further study of vehicle platoon system.</p><p>　　1. Introduction</p><p>　　Reduce the vehicle resistance and the vehicle oil consumption in

46、 some degree. Figure 1 shows the architecture of vehicle platoon system based on vehicle-road cooperation, </p><p>　　which show that the intelligent vehicle control, the vehicle-road information interactive

47、technology, the control way of vehicle platoon and other key technologies </p><p>　　are the important parts[3]However, at present the vehicle platoon's architecture, the behavioral traits and intelligent

48、 travel control algorithm have not been consummated. Therefore, it is necessary to research the foundation, including the behavioral traits of vehicle platoon, modeling of vehicle platoon, and the control method of vehic

49、le platoon. These research need to be conducted by constructing the simulation system of vehicle platoon with hardware in the loop.</p><p>　　For the realization of the vehicle platoon on simulated, intellige

50、nt control and information exchanged is needed. The system needs intelligent control, interactive information, self-government stability. Figure 2 is the system structure of the autonomous smart car. The autonomous smart

51、 car takes MCU as the controller, and uses camera and ultrasonic sensor as navigation sensors, Zigbee as the wireless communication module.</p><p>　　This paper will analyze the overall design of the autonomo

52、us smart car based on the functional require-ments first, then introduce the lateral and longitudinal control and the implementation of the hardware of navigation and the way how to dealt with the key issues, and also di

53、scuss how to improve the car's stability and intelligence level through optimizing the control algori-thms and software, subsequently present its autonomy strategy based on the actual process of testing the autonomou

54、s smar</p><p>　　2.Overall design of autonomous smart simulant car</p><p>　　The autonomous smart car includes four parts: the detection system, the power system, communication system, control and

55、 decision-making system. The detection system is the most important, whose main work is the navigation. We choose CMOS camera as the detection sensor which can detect lane and guide vehicles to travel smoothly on the lin

56、e. In view of a single CMOS image sensor can’ t detect the vehicle distance, overtaking distance and other movement parameters ideal, we choose the ultrasonic se</p><p>　　3. Autonomous smart car hardware sys

57、tem design</p><p>　　Hardware design has a direct impact on the operating effect of the autonomous smart car .According to the above analysis of the autonomous smart car, hardware should have following modul

58、es: (1) navigation units including digital CMOS camera and ultrasonic barrier detected sensor;(2) control units including driving and steering motor; (3) Zigbee wireless communication module; (4) MCU control module.</

59、p><p>　　3.1. Navigation Unit</p><p>　　(1) Camera module design Using camera as a path detection sensor scans the </p><p>　　front path in advance, so that the car can operate more mooth

60、ly. As CMOS image sensor has high integration, low power consumption, low pixel defects and other </p><p>　　advantages, we select OmniVision’s multicolor CMOS image sensor OV6620 with a resolution of 356 * 2

61、92 pixels. Figure 4 is the process of image acquisition. First, MCU controls COMS camera gathering information, then transfer the image data to the buffer memory FIFO, transform the parallel and serial data, finally read

62、 the data </p><p>　　by the MCU’s SPI port.</p><p>　　Image acquisition process has two modes: power up mode and SCCB mode. The system uses SCCB mode: After SCCB initializing OV6620 and enabling V

63、SYNC, the system judge whether it has obtained a frame image. After FIFO stored a frame image, the system gets the data by MCU.</p><p>　　(2) the Design of Ultrasonic Module</p><p>　　If the intel

64、ligent car automatically avoids barrier and navigates, it needs to establish the distance measurement system of the moving vehicle[8]. The ultrasonic distance measurement system can avoid obstacles and locate them, make

65、decision level fusion of information with the camera, and assist path planning. But a small quantity of ultrasonic sensors can't meet the high precision requirements of distance measurement and obstacle avoidance. In

66、 order to ensure the high precision, the car need to</p><p>　　angle limitation of sensor. The system uses 8 ultrasonic measurement channels, using angle compensation means to make location and distance infor

67、mation of ambient </p><p>　　main obstacles more accurately. Figure 5 is installing structure, Figure 6 is the data flow structure diagram of the distance measurement system.</p><p>　　The system

68、calculates distance by detecting time interval from emission to return. Because the time is proportional to ultrasonic distance, when the ultrasonic </p><p>　　transmitting terminal sends several oscillating

69、impulses, MCU begin timing; when the receiver receives first feedback pulse, timing stop. Measured distance as follow: </p><p>　　D=CT/2 (1)</p><p>　　On formula (1), C is sound speed in air

70、, T is the time interval from emission to return.</p><p>　　3.2. Control Unit </p><p>　　(1) DC Motor Drive and Speed Detection </p><p>　　RS-380S-type DC motor is used for speed cont

71、rol. The autonomous smart car takes the closed-loop control technique, and use MC33886 H-bridge driver IC as a </p><p>　　motor driver. If the motor uses open-loop control, it will be subject to many disturbi

72、ng, such as battery voltage, the electrical transmission friction, road friction and forward resistance caused by the front wheel steering angleThese factors will cause operation instability of the smart cars. So closed-

73、loop control method is in urgent need. The closed-loop control system measures speed and adopts </p><p>　　PID algorithm, which needs to acquire the speed changes in a very short interval and calculate the di

74、fference between instantaneous speed and desired speed. Speed </p><p>　　sensor uses OMRON E6A2-CWZ3C encode whose accuracy is up to 360 P / R. </p><p>　　(2) Steer Motor Control </p><

75、p>　　Steer motor is controlled by directly changing the PWM duty cycle of input to change the different turning point[10]. The output angle of the steer motor has a linear </p><p>　　relationship with the P

76、WM signal given. Due to the shift torque of the motor is large enough, MCU calculates the lateral control strategy, and directly gives the PWM </p><p>　　control signals to make realizes the motor shift.</

77、p><p>　　3.3. Wireless communication module </p><p>　　The autonomous smart car’s communications system includes a satellite-based network of ZigBee commu-nications, which consists of a network coord

78、inator and a </p><p>　　number of network terminal nodes. Network coordinator is responsible for the management of the network, while terminal nodes acquire simulated data on the one hand; on the other hand,

79、these simulated data are transmitted through a wireless network to the coordinator. By this way it not only reduces the complexity of the ZigBee network, but also facilitates the centralized management of data. Figure 7

80、describes the communications network effect in the design scheme: within the PAN wireless covera</p><p>　　XDP512 and MC13192. .</p><p>　　3.4. MCU and Peripheral Circuit Module</p><p&g

81、t;　　XDP512 is the core of the autonomous smart car, it controls all the other modules, and acquires path, speed, and wireless signals and other data, on the basis of which </p><p>　　it also classifies the pa

82、rameters and calculates the optimal control strategy. Therefore, the system must have a very high stability. In order to enhance the stability of MCU, the main measures of the design have been taken as follow: ① </p&

83、gt;<p>　　MCU power circuit design; ②filter circuit optimization; ③ MCU system PCB layout of the board; ④ MCU clock Circuit. </p><p>　　4. The software system design of autonomous smart car and relevant

84、 algorithm </p><p>　　The smart car system software consists of the following modules: initialization module, real-time path detection module, anti-collision module, lateral and </p><p>　　longitu

85、dinal closed-loop control module, and communication module. The flow chart of the system software is shown in Figure 8. </p><p>　　In most of the time, MCU is processing data and images. So highly-efficient i

86、mage processing algorithms and closed-loop control algorithm can save the MCU’s </p><p>　　CPU resource and improve the response speed of the autonomous smart car and it’s whole performance. Path extraction a

87、lgorithms and movement feedback control </p><p>　　algorithms after image acquisition are as follow. </p><p>　　4.1.Path Extraction Algorithms</p><p>　　The path is the edge of the lin

88、e of target detection guidelines[11].Algorithm is: a setting threshold of gray-scale image, for each row in the two-dimensional matrix, top-down two adjacent pixels obtained the difference </p><p>　　between

89、the values of (the cut). If the margin is greater than or equal the threshold, its next point of the corresponding pixel is the edge of the guidelines,and the </p><p>　　pixel is considered as the feature poi

90、nt, at the same time, record it. When finding the edge of arrange, we can find out the edge of next line near this line, so take less time to find the point. The algorithm can always track in each column on the edge of

91、the nearby, and find out the edge of next row, So it is high efficient. In lateral control, we make the motor shift according to the coordinates. The look-up table methods can control Lateral direction。</p><p&

92、gt;　　4.2.Controlling algorithms of movement feedback</p><p>　　In longitudinal control, we set a two-dimensional array have 10*33 elements, each row corresponds to a speed value; each arrange corresponds to

93、different corner under </p><p>　　some speed. </p><p>　　In the course of longitudinal control, we set safe speed value according to the current speed and road conditions. The so-called safe speed

94、 is that the car can run cross corners at the speed. When speed feedback is smaller than Speed-Set, car accelerate the speed, if Speed-Feedback equal Speed-Set, car maintain the status and gesture, otherwise, slow down

95、or speed up. Acceleration and deceleration algorithm take as the formula (2): </p><p>　　u(k)=p[e(k)-e(k-1)]+Kie(k)+Kd[e(k)-2e(k-1)+e(k-2)]</p><p>　　△u(k) is increment, which stands for the varia

96、ble of velocity，e(t) is the control error，Kp is proportional gain，Ki = KpT/Ti is integral coefficient，Kd = KpTd/T</p><p>　　is derivative coefficient, Ti is integral const time, Td is derivative const time,

97、and T is the sample time.</p><p>　　5. Test of the autonomous smart car and analysis</p><p>　　As figure 9 shows the appearance of smart car.We test it’s performance from 4 aspects.</p><

98、;p>　　5.1.Patrol line of smart car </p><p>　　After adjustment, the function that autonomous smart car can patrol line came true. It accelerates and decelerates automatically. Computer simulates its traject

99、ory map, figure 10, we can see that car moves based line, and there is certain lateral error. The analysis found that data of lateral control is not detailed enough, then direction control is not meet the control precisi

100、on.</p><p>　　5.2.Test of car avoid obstacle</p><p>　　When the autonomous smart car needs to change lane or overtake, it can avoid obstacles automatically with shock in the process of obstacle-av

101、oiding experiment. The analysis shows that the strategy about avoiding obstacle does not handle the relation of velocity, distance and turning angle well. An accurate model should be founded among them.</p><p&

102、gt;　　5.3.Test of single smart car velocity control</p><p>　　Change the value of Kp,Ki,Kd in the formulate (2) and measure the relationship of motor control and time when smart car is running in high speed, t

103、he relationship </p><p>　　see figure 11. X-coordinate is the measure cycle and y-coordinate is the impulse of measured speed, curve 1 indicates objectspeed, curve 2 and curve 3 are the speed respectively wh

104、en Kp, ki, kd is changed. In experiment, we find Kp influence the system mostly. Figure 11 shows that different amplitude vibration will occur due to different </p><p>　　value of PID parameters in the proces

105、s of speed adjustment. Especially when the speed decline in the response, more steady-state error. </p><p>　　5.4.Test of anti-disturbing and test of communications</p><p>　　Under normal circumst