陳學(xué)深,熊悅淞,齊 龍,王宣霖,程 楠,梁 俊,劉善健
基于觸感引導(dǎo)的小型水田行進(jìn)底盤(pán)自動(dòng)對(duì)行方法
陳學(xué)深,熊悅淞,齊 龍※,王宣霖,程 楠,梁 俊,劉善健
(華南農(nóng)業(yè)大學(xué)工程學(xué)院,廣州 510642)
為了解決小型水田底盤(pán)因路徑偏差導(dǎo)致的稻苗碾壓損傷問(wèn)題,該研究提出一種基于觸感引導(dǎo)的自動(dòng)對(duì)行方法。采用自制的感測(cè)器獲取稻株定位歷程觸感數(shù)據(jù),通過(guò)數(shù)據(jù)的分割閾值設(shè)定、區(qū)域谷值提取、橫向距離標(biāo)定獲得感測(cè)器與稻株的橫向距離。根據(jù)水稻機(jī)械化移栽行距規(guī)整性,利用行距與定位數(shù)據(jù)幾何關(guān)系校驗(yàn)稻株定位數(shù)據(jù),解算獲得稻列方向相鄰稻株中點(diǎn)位置,實(shí)現(xiàn)對(duì)行目標(biāo)點(diǎn)坐標(biāo)提取?;跁r(shí)變坐標(biāo)系跟蹤方法,控制轉(zhuǎn)向電機(jī)實(shí)時(shí)校正路徑偏差,實(shí)現(xiàn)小型水田底盤(pán)自動(dòng)對(duì)行。田間性能試驗(yàn)表明:當(dāng)行進(jìn)速度為0.5 m/s時(shí),自動(dòng)對(duì)行絕對(duì)誤差平均值為3.11 cm、絕對(duì)誤差標(biāo)準(zhǔn)差為1.10 cm、絕對(duì)誤差最大值為4.75 cm,研究成果為水田環(huán)境作業(yè)底盤(pán)自動(dòng)導(dǎo)航提供了新思路和借鑒。
水稻;感測(cè)器;自動(dòng)化;觸覺(jué);定位;對(duì)行
中國(guó)水稻種植區(qū)域地塊分散,有相當(dāng)一部分稻區(qū)處在山區(qū)、丘陵地帶[1-3],大中型植保水田底盤(pán)作業(yè)、轉(zhuǎn)移和運(yùn)輸受到限制[4-6],研究與推廣小型水田底盤(pán)對(duì)植保作業(yè)十分必要。然而,實(shí)際中機(jī)插秧稻田苗帶并非理想的筆直[7-8],小型水田底盤(pán)依靠衛(wèi)星導(dǎo)航[9-10]、視覺(jué)引導(dǎo)[11-12]、人工遙控[13-14]的對(duì)行感知方法存在導(dǎo)航精度差、背景干擾大、視線(xiàn)障礙等問(wèn)題[15]。因此,面向稻田特殊環(huán)境,研究適宜小型水田底盤(pán)自動(dòng)對(duì)行的引導(dǎo)方法具有重要意義。
精準(zhǔn)、實(shí)時(shí)的識(shí)別定位稻株是引導(dǎo)水田底盤(pán)校正路徑偏差,實(shí)現(xiàn)自動(dòng)對(duì)行的關(guān)鍵。國(guó)內(nèi)外學(xué)者主要采用機(jī)器視覺(jué)和機(jī)器觸覺(jué)2種感知方式。其中,視覺(jué)感知又可分為稻株個(gè)體和群體兩種定位方法。關(guān)于稻株個(gè)體定位,Choi等[16]基于形態(tài)學(xué)特征提取水稻冠層植株及葉片骨架的收斂性與延伸性,獲取稻株個(gè)體中心區(qū)域位置。蔣郁等[17]采用側(cè)位俯拍方式獲取了稻株莖基部圖像,基于莖基部分區(qū)邊緣擬合方法實(shí)現(xiàn)了稻株定位。上述稻株個(gè)體定位準(zhǔn)確性高度依賴(lài)特征選擇的有效性,當(dāng)草、藻、萍等稻位俯拍方式獲取了稻株莖基部圖像,基于莖基部分區(qū)田干擾背景特征發(fā)生變化時(shí),稻株識(shí)別定位精度可靠性較差。為提高抗干擾能力,有學(xué)者基于深度學(xué)習(xí)方法自動(dòng)提取稻株特征。Liu等[18]采用VGG16-SSD模型,Lin等[19]基于FasterR-CNN模型實(shí)現(xiàn)了稻株個(gè)體的識(shí)別定位,但由于訓(xùn)練樣本是以稻株個(gè)體為標(biāo)注對(duì)象,當(dāng)?shù)局晗嗷ソ舆B時(shí),樣本特征畸變識(shí)別性能顯著下降,導(dǎo)致稻株定位存在不確定性。
關(guān)于苗帶群體定位,王姍姍等[20]為了有效解決雜草密度分布、光照強(qiáng)度和秧苗行曲率變化等因素對(duì)秧苗行檢測(cè)的影響,提出一種基于特征點(diǎn)鄰域的Hough變換算法識(shí)別秧苗行中心線(xiàn),具有較好的魯棒性和識(shí)別精度。王愛(ài)臣等[21]提出一種基于區(qū)域生長(zhǎng)和均值漂移聚類(lèi)的苗期作物行提取方法,通過(guò)最小二乘法擬合聚類(lèi)中心點(diǎn)得到水稻行直線(xiàn)。張勤等[22]基于窗口掃描法提取了HSI彩色空間中的S分量,在雜草、藍(lán)藻等水田背景下獲取了水稻群體苗帶位置信息。然而,在實(shí)際生產(chǎn)中,稻株群體定位方法易受倒伏、漏插、偏植等實(shí)際情況影響,定位精度較差,且作物行群體特征信息處理量較大,加之復(fù)雜的擬合算法定位實(shí)時(shí)性較低,基于苗帶引導(dǎo)的對(duì)行性能難以滿(mǎn)足要求。
機(jī)器觸覺(jué)是另一種感知方式,在手機(jī)屏幕、機(jī)器人感知手、人造感知皮膚等領(lǐng)域有廣泛應(yīng)用,具有精度高、實(shí)時(shí)性好的特點(diǎn)[23-25]。在農(nóng)業(yè)領(lǐng)域賈洪雷等[26]根據(jù)玉米莖稈力學(xué)閾值,設(shè)計(jì)了一種軟軸式觸覺(jué)傳感器用于識(shí)別定位玉米。徐麗明等[27]根據(jù)葡萄藤力學(xué)參數(shù),設(shè)計(jì)了一種基于臨界壓力值的觸覺(jué)傳感器,獲取了葡萄植株位置。然而,水稻移栽取秧量、生長(zhǎng)分蘗數(shù)不同,難以準(zhǔn)確提取稻株力學(xué)閾值,導(dǎo)致稻株識(shí)別定位可靠性較差。針對(duì)上述問(wèn)題,Chen等[28]提出了一種觸感方法,通過(guò)融合多種稻株觸感特征,在實(shí)驗(yàn)室環(huán)境下實(shí)現(xiàn)了稻株的準(zhǔn)確識(shí)別。Chen等[29]提出了一種用于田間的觸覺(jué)傳感器,基于觸感數(shù)據(jù)特征實(shí)現(xiàn)了稻田雜草密度等級(jí)分類(lèi),但上述觸覺(jué)傳感器僅限于分類(lèi)識(shí)別,不具有稻株定位功能。
面向稻田復(fù)雜環(huán)境,本文提出一種觸感引導(dǎo)的自動(dòng)對(duì)行方法,通過(guò)自制觸覺(jué)感測(cè)器獲取對(duì)行目標(biāo)點(diǎn)坐標(biāo),實(shí)時(shí)校正路徑偏差,擬實(shí)現(xiàn)小型水田底盤(pán)自動(dòng)對(duì)行。
觸感引導(dǎo)系統(tǒng)主要由感測(cè)裝置、轉(zhuǎn)向系統(tǒng)、水田底盤(pán)等組成,總體結(jié)構(gòu)如圖1所示。其中,感測(cè)裝置由左右兩側(cè)具有稻株獨(dú)立定位功能的感測(cè)器,以及單片機(jī)、數(shù)據(jù)傳輸模塊等附屬電子器件組成,為觸感引導(dǎo)提供決策。轉(zhuǎn)向系統(tǒng)采用梯形轉(zhuǎn)向機(jī)構(gòu),由伺服電機(jī)、編碼器、控制器等組成。
1.驅(qū)動(dòng)電池箱 2.電控單元 3.轉(zhuǎn)向電池箱 4.轉(zhuǎn)向伺服電機(jī) 5.右側(cè)感測(cè)器 6.左側(cè)感測(cè)器
工作時(shí),2套感測(cè)器分別位于轉(zhuǎn)向系統(tǒng)前方,隨水田底盤(pán)在水稻行間行進(jìn),通過(guò)接觸實(shí)時(shí)獲取苗帶區(qū)域的稻株觸感數(shù)據(jù),通過(guò)串口傳輸經(jīng)單片機(jī)數(shù)據(jù)處理提取稻株位置信息,并解算稻列方向相鄰稻株中點(diǎn)位置(稻行中心點(diǎn)),實(shí)現(xiàn)對(duì)行目標(biāo)點(diǎn)坐標(biāo)獲取。在此基礎(chǔ)上,控制轉(zhuǎn)向伺服電機(jī)跟蹤目標(biāo)點(diǎn)修正路徑偏差,實(shí)現(xiàn)小型水田底盤(pán)的自動(dòng)對(duì)行。
稻列方向獲取感測(cè)器與稻株的橫向距離,是提取稻株位置,解算對(duì)行目標(biāo)點(diǎn)坐標(biāo)的關(guān)鍵。稻株定位包括觸感數(shù)據(jù)獲取、稻株定位點(diǎn)提取、稻株定位點(diǎn)校驗(yàn)。
采用柔性彎曲傳感器作為觸感數(shù)據(jù)獲取核心元件,利用自身彎曲程度與輸出電壓的對(duì)應(yīng)關(guān)系,獲取接觸、脫離稻株時(shí)段的電壓數(shù)值作為稻株觸感定位的分析數(shù)據(jù)。由于彎曲傳感器自身為薄板懸臂梁結(jié)構(gòu),抗彎剛度小,觸覺(jué)感測(cè)歷程易受稻葉干擾和自身擺振影響。本文對(duì)彎曲傳感器進(jìn)行改制處理,將其疊夾在多片碳纖維薄板之間,在提高自身抗彎剛度的同時(shí),也有利于通過(guò)碳纖維薄板間的相互摩擦抑制自身擺振。另外,考慮水田環(huán)境電子器件的防水保護(hù),彎曲傳感器采用方形鋁管連接固定,并在接線(xiàn)端口處進(jìn)行硅膠密封處理,彎曲傳感器實(shí)物及改制后的稻株觸覺(jué)感測(cè)器實(shí)物如圖2所示。
圖2 彎曲傳感器及觸覺(jué)感測(cè)器實(shí)物圖
水田底盤(pán)行進(jìn)工作時(shí),感測(cè)器水平橫向與稻株莖基部接觸,將感測(cè)器從接觸到脫離稻株歷程作為定位感測(cè)單元,獲取單元內(nèi)感測(cè)器的電壓數(shù)值,為稻株定位提供數(shù)據(jù)基礎(chǔ)。其中,本文設(shè)定的感測(cè)器采樣頻率為1 700 Hz。觸感數(shù)據(jù)獲取實(shí)況如圖3所示,結(jié)構(gòu)參數(shù)如表1所示。
圖3 稻株觸感數(shù)據(jù)獲取實(shí)況
表1 感測(cè)器結(jié)構(gòu)參數(shù)
感測(cè)器與稻株從接觸到脫離歷程,電壓數(shù)值逐漸減小,當(dāng)降低到谷值(最小電壓值)時(shí)恰為感測(cè)器最大彎曲時(shí)刻,而后感測(cè)器回正電壓升高。當(dāng)感測(cè)器根部與稻株接觸時(shí),脫離時(shí)刻彎曲程度最大,中部接觸次之,端部接觸最小。因此,準(zhǔn)確提取感測(cè)器脫離稻株時(shí)刻的電壓值,即觸感歷程電壓數(shù)據(jù)波形的谷值可映射感測(cè)器與稻株的橫向距離,實(shí)現(xiàn)稻株位置獲取。
稻株定位點(diǎn)提取的數(shù)據(jù)處理過(guò)程主要包括分割閾值設(shè)定、區(qū)域谷值提取、橫向距離標(biāo)定。分割閾值設(shè)定是為了提取感測(cè)單元內(nèi)有效的觸感數(shù)據(jù),過(guò)濾掉水稻株間非接觸區(qū)域和感測(cè)器自身擺振形成的冗余數(shù)據(jù)。根據(jù)感測(cè)器擺振數(shù)據(jù)波動(dòng)隨行進(jìn)速度增加而增大特性,本文選擇水田底盤(pán)最高行進(jìn)速度1.5 m/s進(jìn)行數(shù)據(jù)波形分析,如圖4所示。其中,包括株間非接觸的擺振區(qū)和稻株接觸歷程的感測(cè)區(qū)。擺振區(qū)感測(cè)器電壓在4.10 V上下波動(dòng),感測(cè)區(qū)內(nèi)感測(cè)器電壓先經(jīng)歷顯著下降階段,在脫離稻株時(shí)刻達(dá)到谷值,而后感測(cè)器回正電壓經(jīng)歷上升階段,最后再次進(jìn)入擺振區(qū)。由圖4可知,擺振區(qū)電壓數(shù)值較高,且較為穩(wěn)定,定位點(diǎn)與擺振區(qū)的電壓數(shù)值落差十分顯著。為此,本文稻株定位波形分割閾值設(shè)定為4.08 V,可過(guò)濾掉行進(jìn)速度不超過(guò)1.5 m/s的擺振冗余數(shù)據(jù)。然而,稻田地況及環(huán)境復(fù)雜,當(dāng)水田底盤(pán)姿態(tài)變化加劇,可能導(dǎo)致分割閾值不準(zhǔn)確。因此,稻株定位點(diǎn)獲取還需要進(jìn)一步校驗(yàn)。
圖4 稻株定位數(shù)據(jù)波形
區(qū)域谷值提取是為獲取感測(cè)區(qū)的電壓谷值,實(shí)現(xiàn)稻株定位點(diǎn)初步提取。根據(jù)稻株觸感定位歷程的數(shù)據(jù)變化規(guī)律,以小于定位波形分割閾值時(shí)刻為始點(diǎn),大于閾值時(shí)刻為終點(diǎn),提取感測(cè)單元內(nèi)最小電壓值,提取方法的效果如圖5所示,交替水平線(xiàn)填充區(qū)域?yàn)閿[振形成的電壓數(shù)據(jù)分布范圍,屬于數(shù)據(jù)剔除區(qū)域;點(diǎn)線(xiàn)填充區(qū)域?yàn)楦袦y(cè)單元的電壓數(shù)據(jù)分布范圍,屬于數(shù)據(jù)分割區(qū)域,每個(gè)分割區(qū)內(nèi)波形數(shù)據(jù)的谷值為稻株定位點(diǎn),即感測(cè)器脫離稻株時(shí)刻對(duì)應(yīng)的電壓數(shù)值。
注:為谷值(定位點(diǎn));為數(shù)據(jù)剔除區(qū)域;為數(shù)據(jù)分割區(qū)域。
感測(cè)區(qū)谷值為電壓數(shù)據(jù),需經(jīng)過(guò)感測(cè)器標(biāo)定轉(zhuǎn)換為稻株與感測(cè)器的橫向距離。根據(jù)感測(cè)器彎度程度與輸出電壓呈線(xiàn)性反比關(guān)系[30]。標(biāo)定時(shí),感測(cè)器直線(xiàn)行進(jìn)接觸稻株,輸出電壓隨感測(cè)器彎度增大而逐漸減小,當(dāng)脫離稻株時(shí)刻,感測(cè)器回正使輸出電壓驟然增大,記錄此歷程感測(cè)器的電壓谷值,并同時(shí)記錄脫離稻株時(shí)刻感測(cè)器與稻株的橫向距離,將電壓與橫向距離進(jìn)行數(shù)值關(guān)聯(lián),經(jīng)過(guò)多組記錄、測(cè)量、關(guān)聯(lián)和數(shù)據(jù)擬合處理,建立的電壓與橫向距離映射關(guān)系如圖6所示。
通過(guò)感測(cè)器與稻株莖基部邊緣的橫向距離人工測(cè)量值和感測(cè)器測(cè)定值比較,獲得左右兩側(cè)感測(cè)器最大測(cè)量絕對(duì)誤差分別為0.54、0.60 cm,平均測(cè)量絕對(duì)誤差分別為0.28、0.27 cm。在稻株與感測(cè)器相對(duì)位置確定基礎(chǔ)上,結(jié)合感測(cè)器相對(duì)于水田底盤(pán)車(chē)身中心線(xiàn)的相對(duì)位置,可獲得感測(cè)器測(cè)定的橫向距離,即稻株與車(chē)身中心線(xiàn)的橫向距離。
注:L1、L2分別為左、右側(cè)感測(cè)器與稻株的橫向距離,cm;V1、V2分別為左、右側(cè)感測(cè)器的輸出電壓,V。
定位歷程的分割閾值設(shè)定、區(qū)域谷值提取可能因漏插、偏植,以及感測(cè)器自身擺振等情況存在不確定性,形成稻株的偽定位點(diǎn),影響稻株定位的可靠性。因此,在左右兩側(cè)感測(cè)器雙定位基礎(chǔ)上,根據(jù)農(nóng)藝特點(diǎn),充分利用水稻機(jī)械化種植行距規(guī)整性校驗(yàn)稻株定位點(diǎn)。
稻株定位點(diǎn)校驗(yàn)原理如圖7所示,理論上左側(cè)感測(cè)器測(cè)定的橫向距離與右側(cè)感測(cè)器測(cè)定的橫向距離之和等于水稻行距(約為30 cm)。然而,稻株移栽取秧量、生長(zhǎng)分蘗數(shù)不同導(dǎo)致水稻株穴直徑存在差異,加之感測(cè)器自身測(cè)量誤差,實(shí)際作業(yè)中,左右兩側(cè)感測(cè)器測(cè)定的橫向距離之和只能接近水稻行距。考慮水稻株穴半徑差異約為2 cm,水稻行距范圍應(yīng)為26~34 cm。根據(jù)2.2節(jié)所述的左右兩側(cè)感測(cè)器平均測(cè)量絕對(duì)誤差約為0.3 cm,本文基于水稻行距校驗(yàn)閾值設(shè)定為0.6 cm(左右兩側(cè)感測(cè)器平均測(cè)量絕對(duì)誤差之和),即左右兩側(cè)感測(cè)器測(cè)定的橫向距離之和應(yīng)在25.4~34.6 cm(行距±校驗(yàn)閾值)之間。否則,校驗(yàn)判定為稻株偽定位點(diǎn)。經(jīng)過(guò)定位數(shù)據(jù)校驗(yàn),提高了稻株定位的可靠性,為對(duì)行目標(biāo)點(diǎn)的準(zhǔn)確提取提供了基礎(chǔ)。
注:d1、d2分別為左、右側(cè)稻株與車(chē)身中心線(xiàn)的橫向距離,cm;d為水稻行距,cm。
應(yīng)用簡(jiǎn)化二輪車(chē)數(shù)學(xué)模型[31],提出一種基于時(shí)變坐標(biāo)系的跟蹤方法。對(duì)行目標(biāo)點(diǎn)跟蹤原理如圖8所示,以水田底盤(pán)后輪軸中心作為參考原點(diǎn),底盤(pán)行進(jìn)方向?yàn)檩S,垂直行進(jìn)方向?yàn)檩S,建立動(dòng)態(tài)坐標(biāo)系。
稻列方向相鄰稻株的中點(diǎn)作為對(duì)行目標(biāo)點(diǎn),根據(jù)感測(cè)器安裝位置及水田底盤(pán)車(chē)身結(jié)構(gòu)尺寸,本文目標(biāo)點(diǎn)縱坐標(biāo)為定值105 cm。
注:A為對(duì)行目標(biāo)點(diǎn);x為目標(biāo)點(diǎn)與水田底盤(pán)后輪軸中心的橫向距離,cm;y為感測(cè)器與水田底盤(pán)后輪軸中心O的縱向距離,cm;O1為水田底盤(pán)轉(zhuǎn)向的瞬時(shí)圓心;R為底盤(pán)給定轉(zhuǎn)向角情況下所遵循的圓軌跡半徑,cm;L為底盤(pán)前后軸之間的距離,cm;Ld為目標(biāo)點(diǎn)到后輪軸中心的距離,cm;δ為轉(zhuǎn)向角,(°);θ為目標(biāo)點(diǎn)與后輪軸中心的夾角,(°)。
由圖8幾何關(guān)系可獲得式(1)、(2)。
在△l中,由正弦定理可得:
將式(4)代入式(1)中可得:
將式(2)代入式(5)中可得:
由式(6)可知,水田底盤(pán)前后軸之間的距離,以及感測(cè)器與水田底盤(pán)后輪軸中心的縱向距離為定值,轉(zhuǎn)向角由目標(biāo)點(diǎn)橫坐標(biāo)唯一確定,通過(guò)左右兩側(cè)感測(cè)器實(shí)時(shí)獲取稻株位置,并解算對(duì)行目標(biāo)點(diǎn)橫坐標(biāo),實(shí)現(xiàn)水田底盤(pán)路徑偏差校正。
目標(biāo)點(diǎn)橫坐標(biāo)等于左右兩側(cè)感測(cè)器測(cè)定橫向距離之差。因左右兩側(cè)感測(cè)器脫離稻株時(shí)刻存在差異,使定位時(shí)刻不同,但定位差異時(shí)間較短,期間水田底盤(pán)航向變化不大,本文目標(biāo)點(diǎn)橫坐標(biāo)獲取忽略稻株定位時(shí)差內(nèi)航向變化的影響。
目標(biāo)點(diǎn)橫坐標(biāo)獲取包括左右兩側(cè)感測(cè)器全部接觸、部分脫離、全部脫離3個(gè)過(guò)程,如圖9所示。全部接觸過(guò)程感測(cè)器獲取稻株定位觸感數(shù)據(jù),當(dāng)一側(cè)感測(cè)器脫離稻株時(shí)刻,通過(guò)數(shù)據(jù)處理提取并記錄感測(cè)器與稻株的橫向距離,待全部脫離時(shí),將左右兩側(cè)感測(cè)器測(cè)定的橫向距離之差作為對(duì)行目標(biāo)點(diǎn)的橫向坐標(biāo),解算方法如式(7)所示。
圖9 目標(biāo)點(diǎn)提取示意圖
=2?1(7)
隨感測(cè)器行進(jìn),對(duì)行目標(biāo)點(diǎn)坐標(biāo)實(shí)時(shí)傳送到電控單元,根據(jù)轉(zhuǎn)向機(jī)構(gòu)與水田底盤(pán)轉(zhuǎn)向角對(duì)應(yīng)關(guān)系,控制轉(zhuǎn)向伺服電機(jī)實(shí)時(shí)校正路徑偏差,實(shí)現(xiàn)小型水田底盤(pán)自動(dòng)對(duì)行。
為驗(yàn)證觸覺(jué)引導(dǎo)的水田底盤(pán)自動(dòng)對(duì)行性能,2022年4月21日,在華南農(nóng)業(yè)大學(xué)水稻試驗(yàn)田進(jìn)行了不同行進(jìn)速度下人工對(duì)行與自動(dòng)對(duì)行的性能比較試驗(yàn)。試驗(yàn)現(xiàn)場(chǎng)如圖10所示。
圖10 試驗(yàn)現(xiàn)場(chǎng)
試驗(yàn)水稻為移栽20 d的雜交稻五優(yōu)1179,水稻行距為30 cm,株距為15 cm,株高為20~35 cm,稻穴株數(shù)為4~12株,稻穴直徑為2~4 cm。田間雜草主要為華南稻區(qū)稗草、千金子等禾本科雜草,株高為3~9 cm。田間積水層厚度為4~8 cm。
不同行進(jìn)速度水田底盤(pán)航向偏差變化、機(jī)身?yè)u擺程度不同,對(duì)行性能存在差異??紤]水田底盤(pán)常規(guī)作業(yè)速度,選擇0.5、1.0、1.5 m/s進(jìn)行人工對(duì)行與自動(dòng)對(duì)行性能對(duì)比試驗(yàn)。
試驗(yàn)前,隨機(jī)選擇3組測(cè)試區(qū),為保證試驗(yàn)條件相同,人工遙控對(duì)行和自動(dòng)對(duì)行的性能試驗(yàn)在苗帶彎度相同區(qū)域進(jìn)行(插秧機(jī)同步作業(yè)區(qū))。調(diào)整水田底盤(pán)位置,使車(chē)輪位于稻行中央,調(diào)節(jié)左右兩側(cè)感知器高度,接觸位置位于稻株莖基部,距積水層8 cm。
試驗(yàn)時(shí),在進(jìn)入測(cè)試區(qū)前保留10 m起步區(qū),人工遙控方式控制水田底盤(pán)的行進(jìn)速度和方向,以保持水田底盤(pán)在測(cè)試區(qū)內(nèi)行進(jìn)速度穩(wěn)定。進(jìn)入測(cè)試區(qū)后,選擇人工對(duì)行和自動(dòng)對(duì)行2種方式分別進(jìn)行性能試驗(yàn)。
試驗(yàn)后,排盡稻田積水,以測(cè)試區(qū)內(nèi)起始稻行中心點(diǎn)為起始零點(diǎn),每隔4個(gè)株距設(shè)定1個(gè)檢測(cè)點(diǎn),共設(shè)定100個(gè)檢測(cè)點(diǎn),由3人分別統(tǒng)計(jì)左右兩側(cè)輪轍印跡中心點(diǎn)與稻行中心點(diǎn)的水平橫向距離,取平均值作為對(duì)行絕對(duì)誤差,若檢測(cè)點(diǎn)有漏插情況,則以上1個(gè)稻株作為檢測(cè)點(diǎn)。同時(shí),選擇絕對(duì)誤差平均值、絕對(duì)誤差標(biāo)準(zhǔn)差、絕對(duì)誤差最大值作為水田底盤(pán)自動(dòng)對(duì)行的性能評(píng)價(jià)指標(biāo)。其中,絕對(duì)誤差平均值和標(biāo)準(zhǔn)差計(jì)算式分別為
式中P為測(cè)試區(qū)內(nèi)水田底盤(pán)對(duì)行絕對(duì)誤差平均值,cm;P為測(cè)試區(qū)內(nèi)對(duì)行絕對(duì)誤差,cm;為測(cè)試區(qū)檢測(cè)點(diǎn)序號(hào);為測(cè)試區(qū)檢測(cè)點(diǎn)總數(shù);SD為水田底盤(pán)對(duì)行絕對(duì)誤差標(biāo)準(zhǔn)差,cm。
不同行進(jìn)速度下,人工對(duì)行與自動(dòng)對(duì)行2種方式的試驗(yàn)性能評(píng)價(jià)指標(biāo)結(jié)果如表2所示。
表2 自動(dòng)對(duì)行與人工對(duì)行試驗(yàn)對(duì)比結(jié)果
由表2可知,水田底盤(pán)行進(jìn)速度在0.5 m/s時(shí),自動(dòng)對(duì)行絕對(duì)誤差平均值、絕對(duì)誤差標(biāo)準(zhǔn)差、絕對(duì)誤差最大值均優(yōu)于人工對(duì)行方式,而隨水田底盤(pán)行進(jìn)速度增加,自動(dòng)對(duì)行性能有所下降,在1.0和1.5 m/s時(shí),人工對(duì)行效果優(yōu)于自動(dòng)對(duì)行方式,但自動(dòng)對(duì)行絕對(duì)誤差最大值為7.97 cm,從水稻行距30 cm,水田底盤(pán)車(chē)輪寬度10 cm的實(shí)際作業(yè)情況看,當(dāng)對(duì)行誤差小于10 cm時(shí)不會(huì)壓傷稻苗。
1)提出了一種觸感引導(dǎo)的自動(dòng)對(duì)行方法,采用觸覺(jué)感知方式獲取對(duì)行目標(biāo)點(diǎn)坐標(biāo),基于時(shí)變坐標(biāo)系跟蹤方法實(shí)時(shí)校正路徑偏差,為稻田環(huán)境下小型水田底盤(pán)自動(dòng)對(duì)行進(jìn)提供了新思路和借鑒。
2)設(shè)計(jì)了一種稻株定位觸覺(jué)感測(cè)器,通過(guò)觸感數(shù)據(jù)獲取、稻株定位點(diǎn)提取、稻株定位點(diǎn)校驗(yàn),實(shí)現(xiàn)了稻株位置的準(zhǔn)確獲取。
3)田間性能比較試驗(yàn)結(jié)果表明:行進(jìn)速度為0.5、1.0、1.5 m/s時(shí),自動(dòng)對(duì)行絕對(duì)誤差最大值分別為4.75、6.81、7.97 cm,均滿(mǎn)足小型水田自走底盤(pán)對(duì)行要求。
[1] 徐春春,孫麗娟,周錫躍,等. 我國(guó)南方水稻生產(chǎn)變化和特點(diǎn)及穩(wěn)定發(fā)展的政策建議[J]. 農(nóng)業(yè)現(xiàn)代化研究,2013,34(2):129-132.
Xu Chunchun, Sun Lijuan, Zhou Xiyue, et al. Characteristics of rice production in south china and policy proposals for its steady development[J]. Research of Agricultural Modernization, 2013, 34(2): 129-132. (in Chinese with English abstract)
[2] 王忠群,梁建,曹光喬,等. 科學(xué)適度發(fā)展南方丘陵山地農(nóng)機(jī)化[J]. 中國(guó)農(nóng)機(jī)化,2011(2):3-8.
Wang Zhongqun, Liang Jian, Cao Guangqiao, et al. Developing agricultural mechanization scientifically and moderately on the hilly lands in the south of China[J]. Chinese Agricultural Mechanization, 2011(2): 3-8. (in Chinese with English abstract)
[3] 劉珍環(huán),李正國(guó),唐鵬欽,等. 近30年中國(guó)水稻種植區(qū)域與產(chǎn)量時(shí)空變化分析[J]. 地理學(xué)報(bào),2013,68(5):680-693.
Liu Zhenhuan, Li Zhengguo, Tang Pengqin, et al. Spatial-temporal changes of rice area and production in China during 1980-2010[J]. Acta Geographica Sinica, 2013, 68(5): 680-693. (in Chinese with English abstract)
[4] 王圖展. 丘陵山區(qū)農(nóng)業(yè)機(jī)械化發(fā)展的制約因素及對(duì)策:以重慶為例[J]. 農(nóng)機(jī)化研究,2013,35(3):24-28.
Wang Tuzhan. The restriction factor and countermeasure of agricultural mechanization in mountain and hill areas: A case study of Chongqing[J]. Journal of Agricultural Mechanization Research, 2013, 35(3): 24-28. (in Chinese with English abstract)
[5] 邵明璽,辛喆,江秋博. 拖拉機(jī)后懸掛橫向位姿調(diào)整的模糊 PID 控制[J]. 農(nóng)業(yè)工程學(xué)報(bào),2019,35(21):34-42.
Shao Mingxi, Xin Zhe, Jiang Qiubo. Fuzzy PID control for lateral pose adjustment of tractor rear suspension[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2019, 35(21): 34-42. (in Chinese with English abstract)
[6] 李云伍,徐俊杰,王銘楓,等. 丘陵山區(qū)田間道路自主行駛轉(zhuǎn)運(yùn)車(chē)及其視覺(jué)導(dǎo)航系統(tǒng)研制[J]. 農(nóng)業(yè)工程學(xué)報(bào),2019,35(1):52-61.
Li Yunwu, Xu Junjie, Wang Mingfeng, et al. Development of autonomous driving transfer trolley on field roads and its visual navigation system for hilly areas[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2019, 35(1): 52-61. (in Chinese with English abstract)
[7] 何杰,朱金光,羅錫文,等. 電動(dòng)方向盤(pán)插秧機(jī)轉(zhuǎn)向控制系統(tǒng)設(shè)計(jì)[J]. 農(nóng)業(yè)工程學(xué)報(bào),2019,35(6):10-17.
He Jie, Zhu Jinguang, Luo Xiwen, et al. Design of steering control system for rice transplanter equipped with steering wheel-like motor[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2019, 35(6): 10-17. (in Chinese with English abstract)
[8] 邢赫,張國(guó)忠,韓宇航,等. 雙腔氣力式水稻精量水田直播機(jī)設(shè)計(jì)與試驗(yàn)[J]. 農(nóng)業(yè)工程學(xué)報(bào),2020,36(24):29-37.
Xing He, Zhang Guozhong, Han Yuhang, et al. Development and experiment of double cavity pneumatic rice precision direct seeder[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2020, 36(24): 29-37. (in Chinese with English abstract)
[9] 偉利國(guó),張權(quán),顏華,等. XDNZ630型水稻插秧機(jī)GPS自動(dòng)導(dǎo)航系統(tǒng)[J]. 農(nóng)業(yè)機(jī)械學(xué)報(bào),2011,42(7):186-190.
Wei Liguo, Zhang Quan, Yan Hua, et al. GPS automatic navigation system design for XDNZ630 rice transplanter[J]. Transactions of the Chinese Society for Agricultural Machinery, 2011, 42(7): 186-190. (in Chinese with English abstract)
[10] 張智剛,朱啟明,何杰,等. 基于RTK-GNSS和MEMS 陀螺儀的車(chē)輛航向角測(cè)量技術(shù)[J]. 華南農(nóng)業(yè)大學(xué)學(xué)報(bào),2019,40(5):34-37.
Zhang Zhigang, Zhu Qiming, He Jie, et al. Vehicle heading angle measurement technology based on RTK-GNSS and MEMS gyroscope[J]. Journal of South China Agricultural University, 2019, 40(5): 34-37. (in Chinese with English abstract)
[11] 胡煉,羅錫文,曾山,等. 基于機(jī)器視覺(jué)的株間機(jī)械除草裝置的作物識(shí)別與定位方法[J]. 農(nóng)業(yè)工程學(xué)報(bào),2013, 29(10):12-18.
Hu Lian, Luo Xiwen, Zeng Shan, et al. Plant recognition and localization for intra-row mechanical weeding device based on machine vision[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2013, 29(10): 12-18. (in Chinese with English abstract)
[12] 刁智華,王會(huì)丹,宋寅卯. 基于機(jī)器視覺(jué)的農(nóng)田機(jī)械導(dǎo)航線(xiàn)提取算法研究[J]. 農(nóng)機(jī)化研究,2015,37(2):35-39.
Diao Zhihua, Wang Huidan, Song Yinmao. Research of farm machinery navigation algorithm based on machine vision[J]. Journal of Agricultural Mechanization Research, 2015, 37(2): 35-39. (in Chinese with English abstract)
[13] 齊龍,廖文強(qiáng),馬旭,等. 水田小型除草機(jī)器人平臺(tái)控制系統(tǒng)設(shè)計(jì)與測(cè)試[J]. 吉林大學(xué)學(xué)報(bào)(工學(xué)版),2013,(4):991-996.
Qi Long, Liao Wenqiang, Ma Xu, et al. Design and testing of control system of mini-weeding-robot platform in rice paddy field[J]. Journal of Jilin University (Engineering and Technology Edition), 2013, 43(4): 991-996(in Chinese with English abstract)
[14] 曹如月,李世超,魏爽,等. 基于 Web-GIS 的多機(jī)協(xié)同作業(yè)遠(yuǎn)程監(jiān)控平臺(tái)設(shè)計(jì)[J]. 農(nóng)業(yè)機(jī)械學(xué)報(bào),2017,48(S1):52-57.
Cao Ruyue, Li Shichao, Wei Shuang, et al. Remote monitoring platform for multi-machine cooperation based on web-GIS[J]. Transactions of The Chinese Society for Agricultural Machinery, 2017, 48(S1): 52-57. (in Chinese with English abstract)
[15] 張漫,季宇寒,李世超,等. 農(nóng)業(yè)機(jī)械導(dǎo)航技術(shù)研究進(jìn)展[J]. 農(nóng)業(yè)機(jī)械學(xué)報(bào),2020,51(4):1-18.
Zhang Man, Ji Yuhan, Li Shichao, et al. Research progress of agricultural machinery navigation technology[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(4): 1-18. (in Chinese with English abstract)
[16] Choi K H, Han S K, Han S H, et al. Morphology-based guidance line extraction for an autonomous weeding robot in paddy fields[J]. Computers and Electronics in Agriculture, 2015, 113: 266-274.
[17] 蔣郁,崔宏偉,區(qū)穎剛,等. 基于莖基部分區(qū)邊緣擬合的稻株定位方法研究[J]. 農(nóng)業(yè)機(jī)械學(xué)報(bào),2017,48(6):23-31.
Jiang Yu, Cui Hongwei, Ou Yinggang, et al. Positioning method of rice seedlings based on edge fitting in separated region of stem base[J]. Transactions of the Chinese Society for Agricultural Machinery, 2017, 48(6): 23-31. (in Chinese with English abstract)
[18] Liu F C, Yang Y, Zeng Y M, et al. Bending diagnosis of rice seedling lines and guidance line extraction of automatic weeding equipment in paddy field[J]. Mechanical Systems and Signal Processing, 2020, 142: 106791.
[19] Lin S M, Jiang Y, Chen X S, et al. Automatic detection of plant rows for a transplanter in paddy field using faster R-CNN[J]. IEEE Access, 2020, 8: 2169-3536.
[20] 王姍姍,余山山,張文毅,等. 基于特征點(diǎn)鄰域Hough變換的水稻秧苗行檢測(cè)[J]. 農(nóng)業(yè)機(jī)械學(xué)報(bào),2020,51(10):18-25.
Wang Shanshan, Yu Shanshan, Zhang Wenyi, et al. Detection of rice seedling rows based on Hough transform of feature point neighborhood[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(10): 18-25. (in Chinese with English abstract)
[21] 王愛(ài)臣,張敏,劉青山,等. 基于區(qū)域生長(zhǎng)均值漂移聚類(lèi)的苗期作物行提取方法[J]. 農(nóng)業(yè)工程學(xué)報(bào),2021,37(19):202-210.
Wang Aichen, Zhang Min, Liu Qingshan, et al. Seedling crop row extraction method based on regional growth and mean shift clustering[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2021, 37(19): 202-210. (in Chinese with English abstract)
[22] 張勤,黃小剛,李彬. 基于彩色模型和近鄰法聚類(lèi)的水田秧苗列中心線(xiàn)檢測(cè)方法[J]. 農(nóng)業(yè)工程學(xué)報(bào),2012,28(17):163-171,295-296.
Zhang Qin, Huang Xiaogang, Li Bin. Detection of rice seedlings rows' centerlines based on color model and nearest neighbor clustering algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2012, 28(17): 163-171, 295-296. (in Chinese with English abstract)
[23] 呂明,呂延. 觸摸屏的技術(shù)現(xiàn)狀、發(fā)展趨勢(shì)及市場(chǎng)前景[J]. 機(jī)床電器,2012,39(3):4-7.
Lyu Ming, Lyu Yan. The technology current situation and developing tendency and market of touch-screen[J]. Machine Tool Electric Apparatus, 2012, 39(3): 4-7. (in Chinese with English abstract)
[24] 熊利榮,鄭偉,羅舒豪. 基于觸覺(jué)感知的家禽凈膛機(jī)械手及其控制系統(tǒng)的設(shè)計(jì)[J]. 農(nóng)業(yè)工程學(xué)報(bào),2018,34(3):42-48.
Xiong Lirong, Zheng Wei, Luo Shuhao. Design of poultry eviscerated manipulator and its control system based on tactile perception[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2018, 34(3): 42-48. (in Chinese with English abstract)
[25] 郭小輝,黃英,毛磊東,等. 可穿戴式電子織物仿生皮膚設(shè)計(jì)與應(yīng)用研究[J]. 儀器儀表學(xué)報(bào),2016,37(4):938-944.
Guo Xiaohui, Huang Ying, Mao Leidong, et al. Research on the design and application of wearable electronic fabric bionic skin[J]. Chinese Journal of Scientific Instrument, 2016, 37(4): 938-944. (in Chinese with English abstract)
[26] 賈洪雷,李森森,王剛,等. 中耕期玉米田間避苗除草裝置設(shè)計(jì)與試驗(yàn)[J]. 農(nóng)業(yè)工程學(xué)報(bào),2018,34(7):15-22.
Jia Honglei, Li Sensen, Wang Gang, et al. Design and experiment of seedling avoidable weeding control device for intertillage maize[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2018, 34(7): 15-22. (in Chinese with English abstract)
[27] 徐麗明,于暢暢,劉文,等. 籬架式栽培葡萄株間除草機(jī)自動(dòng)避障機(jī)構(gòu)優(yōu)化設(shè)計(jì)[J]. 農(nóng)業(yè)工程學(xué)報(bào),2018,34(7):23-30.
Xu Linming, Yu Changchang, Liu Wen, et al. Optimal design on auto obstacle avoidance mechanism of intra-row weeder for trellis cultivated grape[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2018, 34(7): 23-30. (in Chinese with English abstract)
[28] Chen X S, Mao Y Y, Ma X, et al. A Tactile method for rice plant recognition based on machine learning[J]. Sensors (Basel, Switzerland), 2020, 20(18): E5135
[29] Chen X S, Mao Y Y, Xiong Y S, et al. Intra-row weed density evaluation in rice field using tactile method[J]. Computers & Electronics in Agriculture, 2022, 193: 106699
[30] 任芳昉,洪成雨,張孟喜,等. 基于Flex無(wú)線(xiàn)傳感技術(shù)的邊坡位移監(jiān)測(cè)研究[J]. 傳感技術(shù)學(xué)報(bào),2019,32(7):968-972.
Ren Fangfang, Hong Chengyu, Zhang Mengxi, et al. Study on slope displacement monitoring based on flex wireless sensing technology[J]. Chinese Journal of Sensors and Actuators, 2019, 32(7): 968-972. (in Chinese with English abstract)
[31] 呂安濤,宋正河,毛恩榮. 拖拉機(jī)自動(dòng)轉(zhuǎn)向最優(yōu)控制方法的研究[J]. 農(nóng)業(yè)工程學(xué)報(bào),2006,22(8):116-119.
Lyu Antao, Song Zhenghe, Mao Enrong. Optimized control method for tractor automatic steering[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2006, 22(8): 116-119. (in Chinese with English abstract)
Design and experiment of the tactile guidance system for the automatic alignment of small paddy moving chassis
Chen Xueshen, Xiong Yuesong, Qi Long※, Wang Xuanlin, Cheng Nan, Liang Jun, Liu Shanjian
(,510642,)
Chassis alignment can mainly include the satellite navigation, visual guidance, and manual remote control in the small paddy. However, some problems have led to the damage to rice seedlings in the alignment process, such as the low navigation accuracy, large background interference, and multiple line of sight obstacles. In this study, a tactile-guided system was proposed for the automatic alignment of small paddy chassis. Firstly, a self-developed tactile sensor was used to acquire the tactile data, in which a flexible bending sensor was used as the core element for the tactile data acquisition. The bending sensor was placed between several carbon fiber sheets, in order to improve the own bending stiffness and also to suppress the own oscillation. Secondly, the data was processed to obtain the localization coordinates of the rice plant. The positive relationship between the degree of bending of the sensor and the output voltage was utilized to obtain the information on the position of the rice plant. Three steps were included the data segmentation threshold setting, region valley extraction, and lateral distance calibration. The valid unit of tactile sensing was extracted to set a segmentation threshold. The useless data was filtered out in the non-contact area, and the interference data from the own slight vibration. The data in the sensing unit was further processed to extract the voltage valley value in the sensing area, in order to achieve the initial extraction of the location point of the rice plant. The localization point data was converted into the lateral distance between the rice plant and the sensor by the sensor mapping. Thirdly, the positioning points of rice plants were calibrated, according to the regularity that the sum of the lateral distance measured by the left sensor and the lateral distance measured by the right sensor was equal to the rice row distance. As such, the pseudo-localization points of rice plants were eliminated to improve the reliability of rice plant positioning. Finally, a tracking mode with time-varying coordinate system was proposed to establish the dynamic coordinate system. A calculation was realized on the lateral distance between the center point of the rice line and the center of gravity of the body (target point transverse coordinate). The vertical coordinate was the longitudinal distance between the sensor and the center of the rear wheel axle of the paddy chassis. The coordinates of the aligned target point were real-time transmitted to the electronic control unit during the whole alignment process. The steering servo motor was controlled to correct the path deviation in real time, according to the correspondence between the steering mechanism and the steering angle of the paddy field chassis, in order to realize the automatic alignment of the small paddy field chassis. The field performance test showed that the automatic alignment was better than the manual remote alignment, when the driving speed was 0.5 m/s. The average absolute error of the automatic alignment was 3.11 cm, the standard deviation of the absolute error was 1.10 cm, and the maximum absolute error was 4.75 cm, when the driving speed was 0.5 m/s. The performance of automatic alignment decreased slightly with the increase of driving speed. Anyway, the performance can fully meet the requirements of chassis alignment in paddy fields. The finding can provide a new idea and reference for the automatic chassis navigation in paddy field environment.
rice; sensor; automation; tactile; position; alignment
10.11975/j.issn.1002-6819.2022.21.002
S147.2
A
1002-6819(2022)-21-0008-08
陳學(xué)深,熊悅淞,齊龍,等. 基于觸感引導(dǎo)的小型水田行進(jìn)底盤(pán)自動(dòng)對(duì)行方法[J]. 農(nóng)業(yè)工程學(xué)報(bào),2022,38(21):8-15.doi:10.11975/j.issn.1002-6819.2022.21.002 http://www.tcsae.org
Chen Xueshen, Xiong Yuesong, Qi Long, et al. Design and experiment of the tactile guidance system for the automatic alignment of small paddy moving chassis[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2022, 38(21): 8-15. (in Chinese with English abstract) doi:10.11975/j.issn.1002-6819.2022.21.002 http://www.tcsae.org
2022-09-03
2022-10-25
國(guó)家自然科學(xué)基金(51575195);廣東省基礎(chǔ)與應(yīng)用基礎(chǔ)研究基金項(xiàng)目(2021A1515010831);廣州市科技計(jì)劃項(xiàng)目(202206010125)
陳學(xué)深,副教授,研究方向?yàn)楝F(xiàn)代農(nóng)業(yè)技術(shù)與智能裝備。Email:chenxs@scau.edu.cn
齊龍,教授,博士生導(dǎo)師,研究方向?yàn)楝F(xiàn)代農(nóng)業(yè)技術(shù)裝備。Email:qilong@scau.edu.cn