羅陸鋒 譚遠(yuǎn)良 盧清華 鄒湘軍
摘 要:研發(fā)水果采摘機(jī)器人對(duì)提高收獲效率、保證果實(shí)品質(zhì)和減輕勞動(dòng)強(qiáng)度具有重要意義,但由于果園環(huán)境的非結(jié)構(gòu)性,使得采摘機(jī)器人極易因目標(biāo)定位不準(zhǔn)、采摘順序不當(dāng)、夾剪位姿不合理等導(dǎo)致果實(shí)碰傷或刮落,造成該損傷的主要原因是防碰損采摘的視覺(jué)認(rèn)知與執(zhí)行機(jī)構(gòu)耦合問(wèn)題尚未得到解決。為梳理水果采摘機(jī)器人防碰損作業(yè)的最新研究進(jìn)展,從防碰損采摘中果實(shí)多維信息(采摘點(diǎn)、果梗位姿、防碰空間包圍體等)的視覺(jué)感知、采摘機(jī)器人的視覺(jué)認(rèn)知與智能防碰損采摘行為規(guī)劃、防損采摘機(jī)構(gòu)設(shè)計(jì)及其行為控制等三方面進(jìn)行了全面綜述和分析,并對(duì)今后需重點(diǎn)解決的核心關(guān)鍵問(wèn)題進(jìn)行總結(jié)和展望,為進(jìn)一步研究和攻克非結(jié)構(gòu)環(huán)境下水果智能防碰損采摘問(wèn)題提供參考和依據(jù)。
關(guān)鍵詞:工業(yè)機(jī)器人技術(shù);采摘機(jī)器人;末端執(zhí)行器;視覺(jué)認(rèn)知;智能規(guī)劃
中圖分類號(hào):TP391.41 文獻(xiàn)標(biāo)志碼:A
文章編號(hào):1008-1542(2018)03-0204-10
保證水果低損、采收及時(shí)是果農(nóng)面臨的實(shí)際問(wèn)題,鮮食水果的收獲僅靠人工采摘,不僅采摘人員的勞動(dòng)強(qiáng)度大,采摘成本也較高。為了提高采摘效率,保證水果品質(zhì)和減輕采摘人員的勞動(dòng)強(qiáng)度,用機(jī)器人代替人工采摘是未來(lái)水果采收的發(fā)展趨勢(shì),也是未來(lái)智能農(nóng)業(yè)機(jī)械的發(fā)展方向[1]。
機(jī)器人在采摘水果之前,需先由視覺(jué)系統(tǒng)對(duì)作業(yè)目標(biāo)和環(huán)境進(jìn)行感知與定位,再以感知定位信息為基礎(chǔ)進(jìn)行防碰損作業(yè)的認(rèn)知計(jì)算與行為規(guī)劃,最后通過(guò)視覺(jué)伺服控制采摘機(jī)構(gòu)執(zhí)行夾剪作業(yè)[2]。水果經(jīng)常簇生、果梗桿莖小且果實(shí)嬌嫩,而且水果生長(zhǎng)位置的隨機(jī)性,使得果實(shí)之間經(jīng)常出現(xiàn)相互重疊、貼靠和遮擋等復(fù)雜情況,如圖1所示。
這些非結(jié)構(gòu)因素使得機(jī)器人執(zhí)行機(jī)構(gòu)在夾剪過(guò)程中極易因視覺(jué)定位不準(zhǔn)、采摘順序規(guī)劃不當(dāng)、進(jìn)給和夾剪位姿不合理導(dǎo)致果實(shí)碰傷或刮落[3]。造成這些損傷的主要原因在于采摘中果實(shí)信息視覺(jué)認(rèn)知不準(zhǔn)確、夾剪機(jī)構(gòu)設(shè)計(jì)或其行為決策不合理。為解決該難題,提高采摘作業(yè)的精準(zhǔn)度和柔性度,需要采摘機(jī)器人視覺(jué)系統(tǒng)與執(zhí)行機(jī)構(gòu)之間具備良好的自主防碰損耦合能力,該能力主要體現(xiàn)在果實(shí)多維信息的視覺(jué)感知(采摘點(diǎn)、果梗位姿、防碰空間包圍體等)、隨機(jī)多目標(biāo)防碰損采摘順序的視覺(jué)認(rèn)知與規(guī)劃、執(zhí)行機(jī)構(gòu)與果串之間適形防碰損夾剪位姿的自主耦合等行為上,最終用機(jī)構(gòu)、視覺(jué)、傳感器等硬件和軟件實(shí)現(xiàn)其運(yùn)動(dòng)和智能行為決策。
國(guó)內(nèi)外針對(duì)采摘機(jī)器人的研究已有40多年的歷史[2]。荷蘭、美國(guó)、日本、丹麥、法國(guó)等國(guó)研究了甜椒[4](圖2 a))、柑橘[5]、黃瓜[6](圖2 b))、西紅柿[7](圖2 c))、草莓[8]等采摘機(jī)器人。中國(guó)農(nóng)業(yè)大學(xué)、南京農(nóng)業(yè)大學(xué)、華南農(nóng)業(yè)大學(xué)、江蘇大學(xué)、浙江大學(xué)、上海交通大學(xué)、西北農(nóng)林科技大學(xué)、東北農(nóng)業(yè)大學(xué)等單位研究了草莓[9](圖2 d))、蘋果[10-11](圖2 e))、黃瓜[12]、茄子[13]、荔枝[14](圖2 f))、獼猴桃[15]等果蔬的采摘機(jī)器人。但到目前為止,商品化實(shí)用的采摘機(jī)器人依然鮮有報(bào)道,究其原因是多方面的[16],其中對(duì)果實(shí)造成損傷嚴(yán)重是主要原因之一。為實(shí)現(xiàn)無(wú)損采摘,國(guó)內(nèi)外學(xué)者圍繞水果視覺(jué)感知[17-19]、路徑規(guī)劃[6]、采摘機(jī)構(gòu)[20-21]、果實(shí)損傷及力學(xué)特性[22-24]等開(kāi)展了大量研究。為梳理水果采摘機(jī)器人防碰損作業(yè)的最新研究進(jìn)展,本文將從果實(shí)多維信息視覺(jué)感知、采摘機(jī)器人的視覺(jué)認(rèn)知與智能規(guī)劃、防損采摘機(jī)構(gòu)及行為控制等3個(gè)方面進(jìn)行闡述。
1 水果防碰損采摘的視覺(jué)認(rèn)知及機(jī)構(gòu)設(shè)計(jì)研究狀況與分析
1.1 果實(shí)多維信息視覺(jué)感知的研究進(jìn)展
為了能準(zhǔn)確獲取水果的視覺(jué)信息,學(xué)者們以水果色彩空間為基礎(chǔ)研究出了多種圖像分割與識(shí)別方法[25-30]。CHAMELAT等[25]提出利用Zemike距和顏色信息對(duì)復(fù)雜環(huán)境下的葡萄進(jìn)行識(shí)別;REIS等[26]基于照片彩色信息在自然環(huán)境下對(duì)葡萄進(jìn)行識(shí)別與定位;田銳等[27]計(jì)算葡萄圖像在RGB顏色空間中樣本值,根據(jù)樣本值對(duì)目標(biāo)進(jìn)行圖像分割;熊俊濤等[28]通過(guò)提取顏色模型YCrCb中Cr分量對(duì)成熟荔枝進(jìn)行識(shí)別;羅陸鋒等[29-30]針對(duì)對(duì)果園葡萄的識(shí)別,提出了基于改進(jìn)人工蜂群優(yōu)化模糊聚類的圖像分割方法和基于多顏色空間和集成學(xué)習(xí)的圖像識(shí)別方法。盡管在水果圖像分割方面,國(guó)內(nèi)外學(xué)者已提出了許多有效方法,但僅僅通過(guò)這些方法還難以實(shí)現(xiàn)對(duì)重疊、貼靠和遮擋等非結(jié)構(gòu)情況下串型水果的逐個(gè)精準(zhǔn)識(shí)別與定位。在對(duì)重疊、貼靠和遮擋果實(shí)的識(shí)別方面,國(guó)內(nèi)外對(duì)蘋果[10,31-34]、番茄[35]、柑橘[36]的報(bào)道較多,大多通過(guò)分析果實(shí)幾何形狀來(lái)設(shè)計(jì)相應(yīng)的識(shí)別算法。例如:利用凸殼算法[32]、圓檢測(cè)[10]、輪廓還原[36]等方法對(duì)蘋果目標(biāo)進(jìn)行識(shí)別;以深度圖為基礎(chǔ)結(jié)合邊緣曲率分析[35]實(shí)現(xiàn)對(duì)重疊、黏連番茄的識(shí)別與定位。這些方法在特定條件下對(duì)單一水果的重疊、貼靠等復(fù)雜情況具有實(shí)用性,但對(duì)于串型水果來(lái)說(shuō),因果串輪廓的不規(guī)則性,上述方法通用性依然非常有限。
在水果采摘點(diǎn)的識(shí)別與定位研究上,國(guó)內(nèi)外學(xué)者針對(duì)蘋果[10]、番茄[35]等水果提出了以形心作為采摘點(diǎn)的定位方法。BULANON等[37]分別采用單目相機(jī)移動(dòng)方法和雙目相機(jī)的立體視覺(jué)方法對(duì)蘋果的果實(shí)采摘點(diǎn)的空間位置計(jì)算進(jìn)行了研究;XIANG等[35]采用雙目立體視覺(jué)對(duì)成熟番茄進(jìn)行定位研究,以番茄的形心為采摘點(diǎn);李斌等[38]利用單目視覺(jué)研究了菠蘿果實(shí)形心點(diǎn)的求取方法。然而,這些以形心作為采摘點(diǎn)的方法對(duì)于以?shī)A剪果梗為采摘方式的串型水果來(lái)說(shuō),實(shí)用性不強(qiáng)。在以?shī)A剪果梗為采摘方式的水果采摘中,熊俊濤等[39]利用Hough圓擬合方法先對(duì)柑橘圖像進(jìn)行分割,再運(yùn)用約束斜率為-0.45~0.45的Hough直線檢測(cè)對(duì)柑橘果梗上的采摘點(diǎn)進(jìn)行搜索;楊慶華等[40]通過(guò)提取葡萄圖像輪廓的外接矩形對(duì)目標(biāo)進(jìn)行識(shí)別與定位;張鐵中等[18]利用草莓圖像重心和果尖點(diǎn)來(lái)對(duì)采摘點(diǎn)進(jìn)行識(shí)別與定位;郭艾俠等[19]設(shè)計(jì)了一種融合Harris與SIFT算法的荔枝果梗采摘點(diǎn)計(jì)算方法;羅陸鋒等[41]根據(jù)葡萄的生長(zhǎng)特點(diǎn),提出了一種基于點(diǎn)線最小距離約束的采摘點(diǎn)定位方法,如圖3所示。
在果梗位姿的視覺(jué)檢測(cè)研究上,INKYU等[42]對(duì)果園環(huán)境下的甜辣椒果梗檢測(cè)進(jìn)行了研究,先利用RGB-D傳感器采集圖像數(shù)據(jù),獲取點(diǎn)云數(shù)據(jù)后使用幾何特征(如表面法線和曲率估計(jì))對(duì)甜椒三維幾何形狀進(jìn)行識(shí)別,結(jié)合HSV顏色特征和PFH(point feature histograms)對(duì)監(jiān)督學(xué)習(xí)方法進(jìn)行訓(xùn)練,最終實(shí)現(xiàn)對(duì)果梗檢測(cè),過(guò)程如圖4所示。
在水果防碰空間包圍體的求解與定位研究上,羅陸鋒等[43]提出一種基于雙目立體視覺(jué)的葡萄包圍體求解與定位方法,如圖5所示。先通過(guò)尋找與重心距離最小的直線來(lái)定位果梗上的采摘點(diǎn);再運(yùn)用圓檢測(cè)法獲取外接矩形區(qū)域內(nèi)果粒的圓心和半徑,求解采摘點(diǎn)和果粒圓心的空間坐標(biāo);最后以采摘點(diǎn)的空間坐標(biāo)為原點(diǎn)構(gòu)建葡萄空間坐標(biāo)系,求解葡萄最大截面,再將該截面繞中心軸旋轉(zhuǎn)360°得到葡萄空間包圍體。但目前該方法只能處理單串葡萄,對(duì)重疊、貼靠和遮擋葡萄串的防碰包圍體定位還需進(jìn)一步深入研究解決。
1.2 采摘機(jī)器人的視覺(jué)認(rèn)知與智能規(guī)劃的研究進(jìn)展
為避免損傷果串,采摘機(jī)器人在作業(yè)規(guī)劃中需要先根據(jù)環(huán)境信息自主規(guī)劃防碰損的采摘順序后再到達(dá)采摘點(diǎn)執(zhí)行夾剪作業(yè);在夾持和剪斷果梗時(shí),又需要機(jī)器人能自主確立防碰損的執(zhí)行機(jī)構(gòu)進(jìn)給方位和夾剪角度。BAC等[44]根據(jù)視覺(jué)定位信息約束末端執(zhí)行器旋轉(zhuǎn)切刀(圖6)的初始方位角來(lái)進(jìn)行防損采摘,從而實(shí)現(xiàn)對(duì)甜椒的低損收獲,但甜椒與串型水果的形態(tài)特征差異較大,且采摘方式亦不盡相同,因而該方法對(duì)于串型水果來(lái)說(shuō)可移植性不強(qiáng);HENTEN等[6]從逆運(yùn)動(dòng)學(xué)角度研究了黃瓜采摘機(jī)械臂的免碰撞路徑規(guī)劃問(wèn)題,逆運(yùn)動(dòng)學(xué)是一個(gè)非線性規(guī)劃問(wèn)題,通過(guò)遺傳算法解決,但是由于過(guò)程需要大量的計(jì)算時(shí)間,所以該黃瓜采摘機(jī)器人的基本模型采用有冗余度的P6R操作器。梁喜鳳等[45]基于偽距離避障法,以番茄收獲機(jī)械手可操作度最大化為目標(biāo)函數(shù),采用偽距離避障法和迭代法相結(jié)合的方法能夠使番茄收獲機(jī)械手在保證良好工作性能的前提下實(shí)現(xiàn)避障運(yùn)動(dòng)規(guī)劃,機(jī)械手由初始位置沿預(yù)定路徑運(yùn)動(dòng)至目標(biāo)位置,并能成功避開(kāi)障礙物;尹建軍等[46]以關(guān)節(jié)型機(jī)械臂避開(kāi)垂直莖稈或撐桿采摘番茄為研究對(duì)象,提出了一種基于構(gòu)形空間的關(guān)節(jié)型機(jī)械臂避障路徑規(guī)劃方法,以能量最優(yōu)函數(shù)優(yōu)選避障規(guī)劃的關(guān)節(jié)終點(diǎn)角,利用A*算法可以得到平面R-R機(jī)械臂的避障路徑,獲得一系列表示空間連桿位置的相交豎直面,并在豎直面內(nèi)進(jìn)行其余關(guān)節(jié)角的規(guī)劃。
在采摘順序規(guī)劃方面,王冰心等[47]基于仿生學(xué)思想設(shè)計(jì)了一種基于視覺(jué)選擇性注意機(jī)制的果實(shí)簇識(shí)別與采摘順序規(guī)劃方法,如圖7所示。該研究一方面有利于提高果實(shí)簇識(shí)別算法對(duì)復(fù)雜農(nóng)業(yè)環(huán)境的適應(yīng)能力,提高算法對(duì)不同種類果實(shí)識(shí)別的通用性;另一方面通過(guò)提前規(guī)劃視域內(nèi)多個(gè)果實(shí)簇的采摘順序,減少采摘過(guò)程中的重復(fù)動(dòng)作,提高作業(yè)效率。但目前該方法尚未考慮采摘過(guò)程中的防碰損規(guī)劃問(wèn)題;羅陸鋒等[48]開(kāi)展對(duì)多目標(biāo)防碰損采摘順序規(guī)劃算法進(jìn)行仿真試驗(yàn)研究,將實(shí)物視覺(jué)與虛擬現(xiàn)實(shí)相結(jié)合設(shè)計(jì)了采摘機(jī)器人硬件在環(huán)虛擬試驗(yàn)系統(tǒng),為防碰損采摘的視覺(jué)定位及行為控制算法提供了仿真試驗(yàn)平臺(tái)。
1.3 防損采摘機(jī)構(gòu)及行為控制的研究進(jìn)展
采摘機(jī)器人采摘機(jī)構(gòu)及行為模型對(duì)實(shí)現(xiàn)無(wú)損采摘至關(guān)重要。BLANES等[49]通過(guò)設(shè)計(jì)具有形狀自適應(yīng)的夾指機(jī)構(gòu)來(lái)降低茄子采摘中的機(jī)械損傷,如圖9、圖10所示,該末端執(zhí)行器由3個(gè)機(jī)械手指和1個(gè)真空吸盤組成,其中3個(gè)機(jī)械手指均裝有慣性傳感器,當(dāng)末端執(zhí)行器接觸不同形狀的茄子時(shí),其內(nèi)部顆粒狀物質(zhì)的干擾從柔軟到堅(jiān)硬,其中1根手指可以適應(yīng)并復(fù)制茄子的形狀,同時(shí)其他的手指也可以使用冗余的自由度適應(yīng)茄子的形狀;劉繼展等[21]針對(duì)機(jī)器人摘取及移送過(guò)程中導(dǎo)致的果穗振動(dòng)與果粒脫落問(wèn)題,提出了一種面向穗軸激勵(lì)輸入的果穗振動(dòng)仿真模型,該研究以葡萄為研究對(duì)象,在果穗“梗-果”結(jié)構(gòu)特性基礎(chǔ)上,提出了“撓性桿-鉸鏈-剛性桿-質(zhì)量球”復(fù)合果穗模型,并由試驗(yàn)確定了模型中各級(jí)梗間鉸鏈彈性系數(shù)與阻尼系數(shù)、果粒尺寸與質(zhì)量的正態(tài)分布規(guī)律,獲得了主穗軸的抗彎特性,進(jìn)而利用激光3D掃描重構(gòu)得到梗系統(tǒng),根據(jù)試驗(yàn)結(jié)果分別進(jìn)行剛性、撓性桿件定義和果粒與梗間鉸鏈的添加,構(gòu)建得到果穗振動(dòng)仿真模型;金波等[20]為了實(shí)現(xiàn)果蔬的無(wú)損采摘,采用欠驅(qū)動(dòng)原理設(shè)計(jì)出具有形狀自適應(yīng)能力的末端執(zhí)行器夾指,如圖11、圖12所示,該手爪用1個(gè)驅(qū)動(dòng)電動(dòng)機(jī)控制3個(gè)手指、9個(gè)指節(jié),可以實(shí)現(xiàn)對(duì)果實(shí)的包絡(luò)抓取,并且通過(guò)PID閉環(huán)力控制方式,實(shí)現(xiàn)了對(duì)遠(yuǎn)指關(guān)節(jié)和所有關(guān)節(jié)中最大接觸力的有效控制,實(shí)現(xiàn)期望的抓取與最大接觸力控制功能,并具有控制簡(jiǎn)單可靠、抓取穩(wěn)定、不損傷果實(shí)等特點(diǎn);姬偉等[22]為減少夾持器抓取蘋果時(shí)的碰撞、擠壓損傷,通過(guò)壓縮試驗(yàn)后計(jì)算得到了蘋果果皮、果肉和果核3個(gè)不同部分的力學(xué)參數(shù),建立了單個(gè)蘋果的3層實(shí)體力學(xué)模型,為蘋果收獲機(jī)器人夾持器結(jié)構(gòu)設(shè)計(jì)和控制方法提供參考數(shù)據(jù)和可控措施;陳燕等[50]為減少采摘中荔枝的機(jī)械損傷,通過(guò)建立采摘器剪切運(yùn)動(dòng)和剪切力模型對(duì)末端執(zhí)行器的結(jié)構(gòu)進(jìn)行了參數(shù)優(yōu)化;李娜等[51]針對(duì)溫室內(nèi)地壟式栽培草莓自動(dòng)低損采摘的需求,利用氣動(dòng)肌腱驅(qū)動(dòng)創(chuàng)新設(shè)計(jì)了多指式剛?cè)峄炻?lián)欠驅(qū)動(dòng)草莓采摘機(jī)械手,如圖13所示,該機(jī)械手設(shè)置欠驅(qū)動(dòng)關(guān)節(jié),具有結(jié)構(gòu)簡(jiǎn)單、低能耗的特點(diǎn),同時(shí)在手指單元中加入柔順構(gòu)件,利用其柔性變形可減小采摘中對(duì)果實(shí)的損壞;饒洪輝等[52]為實(shí)現(xiàn)油茶果低損采摘,設(shè)計(jì)一種氣吸式油茶果采摘機(jī)構(gòu),該機(jī)構(gòu)先通過(guò)真空泵產(chǎn)生的負(fù)壓吸住待摘油茶果,再用電機(jī)驅(qū)動(dòng)真空吸盤旋轉(zhuǎn)來(lái)對(duì)油茶果進(jìn)行旋脫;王學(xué)林等[53]為減少末端執(zhí)行器對(duì)果蔬抓持損傷,設(shè)計(jì)了基于灰色預(yù)測(cè)的增量式比例積分夾持力控制算法,實(shí)現(xiàn)夾持機(jī)構(gòu)和果蔬之間動(dòng)態(tài)抓持過(guò)程的自適應(yīng)調(diào)整;李建偉等[54]根據(jù)目前蘋果采摘末端執(zhí)行器存在的問(wèn)題,設(shè)計(jì)了依靠驅(qū)動(dòng)刀片在動(dòng)力驅(qū)使下繞指外圍圓周方向自由旋轉(zhuǎn)1周,切除蘋果柄任意位置的末端執(zhí)行器,如圖14所示,設(shè)計(jì)方式既可避免刀片傷到蘋果,又可以切斷蘋果柄,極大提高了采摘效率,且控制程序簡(jiǎn)單,降低了機(jī)器成本;羅陸鋒等[55]通過(guò)分析葡萄等串型水果的形狀和生長(zhǎng)特點(diǎn),設(shè)計(jì)了一種夾持-托舉-剪斷式串型水果采摘機(jī)構(gòu),該機(jī)構(gòu)先由柔性?shī)A指夾持住果梗,再用托盤從后下方托舉果串以防脫落,最后通過(guò)剪刀對(duì)果梗進(jìn)行切斷,但目前尚未使用該執(zhí)行器進(jìn)行視覺(jué)關(guān)聯(lián)的采摘試驗(yàn),其機(jī)構(gòu)與控制參數(shù)還有待通過(guò)采摘試驗(yàn)進(jìn)行優(yōu)化和修正。
2 研究展望
通過(guò)國(guó)內(nèi)外研究狀況分析發(fā)現(xiàn):目前針對(duì)機(jī)器人無(wú)損采摘的研究主要集中于果實(shí)目標(biāo)的識(shí)別與定位[17-19,56-57]、免碰撞路徑規(guī)劃[6,46]、柔順夾切機(jī)構(gòu)設(shè)計(jì)[20]、果串防脫落夾持模型[21]、損傷力學(xué)特性[22-24]和采摘中夾持力控制[53]等方面,而對(duì)于非結(jié)構(gòu)環(huán)境下水果采摘機(jī)構(gòu)與視覺(jué)關(guān)聯(lián)的智能行為研究還很少,為增強(qiáng)采摘機(jī)器人的智能化、無(wú)損化作業(yè)水平,未來(lái)還有諸多問(wèn)題需要研究解決。
1)戶外果園環(huán)境下光照強(qiáng)度的不確定性使得水果識(shí)別算法的魯棒性受到了很大挑戰(zhàn),目前還未發(fā)現(xiàn)有通用算法能解決不斷變化光照下的目標(biāo)識(shí)別問(wèn)題,盡管世界各地的學(xué)者從視覺(jué)傳感器、圖像處理算法上進(jìn)行了大量研究,但仍需要進(jìn)一步研究和完善。隨著深度學(xué)習(xí)技術(shù)的不斷發(fā)展,水果識(shí)別算法的魯棒性打破了傳統(tǒng)的圖像識(shí)別方法,目前已獲得極大的成功,該方法通過(guò)構(gòu)建多層深度學(xué)習(xí)的神經(jīng)網(wǎng)絡(luò)來(lái)模擬人腦的認(rèn)知規(guī)律,但需要通過(guò)大量數(shù)據(jù)進(jìn)行卷積神經(jīng)網(wǎng)絡(luò)訓(xùn)練,未來(lái)將深度學(xué)習(xí)運(yùn)用于農(nóng)業(yè)非結(jié)構(gòu)環(huán)境中的目標(biāo)識(shí)別將是一個(gè)非常有潛力的方向。
2)果實(shí)果梗采摘點(diǎn)的識(shí)別與定位是采摘機(jī)器人視覺(jué)系統(tǒng)的核心難點(diǎn)問(wèn)題,盡管國(guó)內(nèi)外學(xué)者已經(jīng)提出了一些有效的實(shí)現(xiàn)方法,但大多數(shù)是針對(duì)單一水果,通用性不強(qiáng),且精準(zhǔn)度還有待進(jìn)一步提升。在今后的研究中,以深度相機(jī)等先進(jìn)視覺(jué)傳感器為基礎(chǔ),綜合運(yùn)用點(diǎn)云形態(tài)學(xué)建模與智能計(jì)算方法來(lái)構(gòu)建果實(shí)識(shí)別模型,可有望提高非結(jié)構(gòu)環(huán)境下果實(shí)采摘點(diǎn)的識(shí)別與定位準(zhǔn)確度。
3)非結(jié)構(gòu)因素使得機(jī)器人在夾剪過(guò)程中極易因定位不準(zhǔn)而導(dǎo)致末端執(zhí)行器損傷水果;容易因采摘順序規(guī)劃不當(dāng)造成果串之間發(fā)生干涉碰撞而導(dǎo)致水果滑落損傷;當(dāng)末端執(zhí)行器靠近采摘點(diǎn)時(shí),又容易因進(jìn)給方向和夾剪角度設(shè)置不恰當(dāng)導(dǎo)致執(zhí)行機(jī)構(gòu)碰傷果實(shí)。在今后的研究中,需將采摘機(jī)器人作業(yè)方式與水果規(guī)范化種植進(jìn)行有機(jī)結(jié)合,實(shí)現(xiàn)農(nóng)機(jī)和農(nóng)藝高度融合,最終用機(jī)構(gòu)、視覺(jué)、傳感器等硬件和軟件實(shí)現(xiàn)機(jī)器人防損采摘的智能行為決策。
4)設(shè)計(jì)具有智能防碰損采摘能力的夾剪執(zhí)行機(jī)構(gòu)及其視覺(jué)伺服系統(tǒng)是采摘機(jī)器人關(guān)鍵所在,傳統(tǒng)采摘機(jī)構(gòu)及其控制系統(tǒng)的設(shè)計(jì)大多依據(jù)采摘對(duì)象的形狀特點(diǎn)及其生物力學(xué)特性來(lái)設(shè)計(jì)相應(yīng)的機(jī)構(gòu)構(gòu)型與控制方法,其視覺(jué)系統(tǒng)和采摘機(jī)構(gòu)通常是分開(kāi)設(shè)計(jì)后再進(jìn)行,在今后的研究中非常需要對(duì)防碰損采摘的視覺(jué)認(rèn)知與執(zhí)行機(jī)構(gòu)進(jìn)行協(xié)同耦合建模,通過(guò)對(duì)耦合過(guò)程模型進(jìn)行反復(fù)試驗(yàn)來(lái)改進(jìn)和優(yōu)化水果防碰損采摘系統(tǒng)的機(jī)構(gòu)設(shè)計(jì)參數(shù)和視覺(jué)伺服控制參數(shù)。
參考文獻(xiàn)/References:
[1] 王儒敬, 孫丙宇. 農(nóng)業(yè)機(jī)器人的發(fā)展現(xiàn)狀與展望[J].中國(guó)科學(xué)院院刊,2015,30(6):803-809.
WANG Rujing, SUN Bingyu. Development status and expectation of agricultural robot[J]. Bulletin of Chinese Academy of Science, 2015,30(6):803-809.
[2] GONGAL A, AMATYA S, KARKEE M, et al. Sensors and systems for fruit detection and localization: A review[J]. Computers and Electronics in Agriculture,2015,116(8):8-19.
[3] BECHAR A, VIGNEAULT C. Agricultural robots for field operations: Concepts and components[J]. Biosystems Engineering, 2016, 149: 94-111.
[4] BAC C W, HEMMING J, HENTEN E J V. Robust pixel-based classification of obstacles for robotic harvesting of sweet-pepper[J].Computers and Electronics in Agriculture, 2013, 96(12):148-162.
[5] MEHTA S S, BURKS T F. Vision-based control of robotic manipulator for citrus harvesting[J]. Computers and Electronics in Agriculture, 2014,102(3):146-158.
[6] HENTEN E J V, SCHENK E J, WILLIGENBURG L G V, et al. Collision-free inverse kinematics of the redundant seven-link manipulator used in a cucumber picking robot[J]. Biosystems Engineering, 2010, 106(2): 112-124.
[7] KONDO N, YATA K, IIDA M, et al. Development of an end-effector for a tomato cluster harvesting robot[J]. Engineering in Agriculture Environment and Food, 2010,3(1):20-24.
[8] HAYASHI S, YAMAMOTO S, SAITO S, et al. Structural environment suited to the operation of a strawberry-harvesting robot mounted on a travelling platform[J]. Engineering in Agriculture Environment and Food,2013, 6(1):34-40.
[9] 王糧局,張立博,段運(yùn)紅,等.基于視覺(jué)伺服的草莓采摘機(jī)器人果實(shí)定位方法[J].農(nóng)業(yè)工程學(xué)報(bào), 2015,31(22):25-31.
WANG Liangju, ZHANG Libo, DUAN Yunhong, et al. Fruit localization for strawberry harvesting robot based on visual servoing[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015,31(22):25-31.
[10]SI Y, LIU G, FENG J. Location of apples in trees using stereoscopic vision[J]. Computers and Electronics in Agriculture, 2015, 112(3): 68-74.
[11]ZHAO D A, LYU J, JI W, et al. Design and control of an apple harvesting robot[J]. Biosystems Engineering, 2011, 110(2): 112-122.
[12]紀(jì)超,馮青春,袁挺,等.溫室黃瓜采摘機(jī)器人系統(tǒng)研制及性能分析[J].機(jī)器人, 2011, 33(6): 726-730.
JI Chao, FENG Qingchun, YUAN Ting, et al. Development and performance analysis on cucumber harvesting robot system in greenhouse[J]. Robot, 2011, 33(6): 726-730.
[13]宋健,孫學(xué)巖,張鐵中,等. 開(kāi)放式茄子采摘機(jī)器人設(shè)計(jì)與試驗(yàn)[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2009, 40(1):143-147.
SONG Jian, SUN Xueyan, ZHANG Tiezhong, et al. Design and experiment of opening picking robot for eggplant[J]. Transactions of the Chinese Society for Agricultural Machinery, 2009, 40(1):143-147.
[14]葉敏,鄒湘軍,羅陸鋒,等.荔枝采摘機(jī)器人雙目視覺(jué)的動(dòng)態(tài)定位誤差分析[J].農(nóng)業(yè)工程學(xué)報(bào), 2016, 32(5): 50-56.
YE Min, ZOU Xiangjun, LUO Lufeng, et al. Error analysis of dynamic localization tests based on binocular stereo vision on litchi harvesting manipulator[J]. Transactions of the Chinese Society of Agricultural Engineering, 2016, 32(5): 50-56.
[15]崔永杰,蘇帥,王霞霞,等.基于機(jī)器視覺(jué)的自然環(huán)境中獼猴桃識(shí)別與特征提取[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2013,44(5):247-252.
CUI Yongjie, SU Shuai, WANG Xiaxia, et al. Recognition and feature extraction of kiwifruit in natural environment based on machine vision[J]. Transactions of the Chinese Society for Agricultural Machinery,2013,44(5):247-252.
[16]HAYASHI S, SHIGEMATSU K, YAMAMOTO S, et al. Evaluation of a strawberry-harvesting robot in a field test[J]. Biosystems Engineering, 2010, 105(2): 160-171.
[17]項(xiàng)榮,應(yīng)義斌,蔣煥煜,等.基于雙目立體視覺(jué)的番茄定位[J].農(nóng)業(yè)工程學(xué)報(bào), 2012,28(5):161-167.
XIANG Rong, YING Yibin, JIANG Huanyu, et al. Localization of tomatoes based on binocular stereo vision[J]. Transactions of the Chinese Society of Agricultural Engineering, 2012,28(5):161-167.
[18]張鐵中,陳利兵,宋健.草莓采摘機(jī)器人的研究:Ⅱ.基于圖像的草莓重心位置和采摘點(diǎn)的確定[J].中國(guó)農(nóng)業(yè)大學(xué)學(xué)報(bào),2005,10(1): 48-51.
ZHANG Tiezhong, CHEN Libing, SONG Jian. Study on strawberry harvesting robot: Ⅱ.Images based identifications of strawberry barycenter and plucking position[J]. Journal of China Agricultural University,2005,10(1): 48-51.
[19]郭艾俠,熊俊濤,肖德琴,等.融合Harris與SIFT算法的荔枝采摘點(diǎn)計(jì)算與立體匹配[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2015,46(12):11-17.
GUO Aixia, XIONG Juntao, XIAO Deqin, et al. Computation of picking point of litchi and its binocular stereo matching based on combined algorithm of Harris and SIFT[J]. Transactions of the Chinese Society for Agricultural Machinery, 2015,46(12):11-17.
[20]金波,林龍賢.果蔬采摘欠驅(qū)動(dòng)機(jī)械手爪設(shè)計(jì)及其力控制[J].機(jī)械工程學(xué)報(bào),2014, 50(19): 1-8.
JIN Bo, LIN Longxian. Design and force control of an underactuated robotic hand for fruit and vegetable picking[J]. Journal of Mechanical Engineering, 2014, 50(19): 1-8.
[21]劉繼展,唐善奇,單帥,等.機(jī)器人采摘葡萄果穗振動(dòng)仿真與試驗(yàn)[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2016,47(5):1-8.
LIU Jizhan, TANG Shanqi, SHAN Shuai, et al. Simulation and test of grape fruit cluster vibration for robotic harvesting[J]. Transactions of the Chinese Society for Agricultural Machinery, 2016,47(5):1-8.
[22]姬偉,李俊樂(lè),楊俊,等.機(jī)器手采摘蘋果抓取損傷機(jī)理有限元分析及驗(yàn)證[J].農(nóng)業(yè)工程學(xué)報(bào),2015,31(5):17-22.
JI Wei, LI Junle, YANG Jun, et al. Analysis and validation for mechanical damage of apple by gripper in harvesting robot based on finite element method[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015,31(5):17-22.
[23]陳燕,蔡偉亮,鄒湘軍,等.荔枝的力學(xué)特性測(cè)試及其有限元分析[J].農(nóng)業(yè)工程學(xué)報(bào),2011,27(12): 358-363.
CHEN Yan, CAI Weiliang, ZOU Xiangjun, et al. Mechanical properties test and finite element analysie for litchi[J]. Transactions of the Chinese Society of Agricultural Engineering, 2011,27(12): 358-363.
[24]蔣志林,陳燕,談建豪,等.荔枝應(yīng)力松弛特性及其損傷的試驗(yàn)研究[J].現(xiàn)代食品科技,2015,31(5):96-101.
JIANG Zhilin, CHEN Yan, TAN Jianhao, et al. Experimental study on stress relaxation characteristics and associated damage litchi[J]. Modern Food Science and Technology, 2015,31(5):96-101.
[25]CHAMELAT R, ROSSO E, CHOKSURIWONG A, et al. Grape detection by image processing[C]//Conference of the IEEE Industrial Electronics Society.[S.l.]:[s.n.],2006: 3697-3702.
[26]REIS M J C S, MORAIS R, PERES E, et al. Automatic detection of bunches of grapes in natural environment from color images[J]. Journal of Applied Logic, 2012, 10(4):285-290.
[27]田銳,郭艷玲.基于機(jī)器視覺(jué)的葡萄自動(dòng)識(shí)別技術(shù)[J].東北林業(yè)大學(xué)學(xué)報(bào),2008,36(11):95-97.
TIAN Rui, GUO Yanling. Automatic identification techniques of grape based on computer vision[J]. Journal of Northeast Forestry University, 2008,36(11):95-97.
[28]熊俊濤,鄒湘軍,陳麗娟,等.基于機(jī)器視覺(jué)的自然環(huán)境中成熟荔枝識(shí)別[J].農(nóng)業(yè)機(jī)械學(xué)報(bào), 2011, 42(9): 162-166.
XIONG Juntao, ZOU Xiangjun, CHEN Lijuan, et al. Recognition of mature litchi in natural environment based on machine vision[J]. Transactions of the Chinese Society for Agricultural Machinery, 2011, 42(9): 162-166.
[29]羅陸鋒,鄒湘軍,楊洲,等.基于改進(jìn)人工蜂群模糊聚類的葡萄圖像快速分割方法[J].農(nóng)業(yè)機(jī)械學(xué)報(bào), 2015, 46(3):23-28.
LUO Lufeng, ZOU Xiangjun, YANG Zhou, et al. Grape image fast segmentation based on improved artificial bee colony and fuzzy clustering[J]. Transactions of the Chinese Society for Agricultural Machinery, 2015, 46(3):23-28.
[30]LUO Lufeng, TANG Yunchao, ZOU Xiangjun, et al. Robust grape cluster detection in a vineyard by combining the AdaBoost framework and multiple color components[J]. Sensors,2016, 16(12): 2098-2117.
[31]趙德安,沈甜,陳玉,等.蘋果采摘機(jī)器人快速跟蹤識(shí)別重疊果實(shí)[J].農(nóng)業(yè)工程學(xué)報(bào),2015,31(2):22-28.
ZHAO Dean, SHEN Tian, CHEN Yu, et al. Fast tracking and recognition of overlapping fruit for apple harvesting robot[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015,31(2):22-28.
[32]宋懷波,張傳棟,潘景朋,等.基于凸殼的重疊蘋果目標(biāo)分割與重建算法[J].農(nóng)業(yè)工程學(xué)報(bào),2013,29(3):163-168.
SONG Huaibo, ZHANG Chuandong, PAN Jingpeng, et al. Segmentation and reconstruction of overlapped apple images based on convex hull[J]. Transactions of the Chinese Society of Agricultural Engineering, 2013,29(3):163-168.
[33]WANG D, SONG H, TIE Z, et al. Recognition and localization of occluded apples using K-means clustering algorithm and convex hull theory: A comparison[J]. Multimedia Tools and Applications, 2016,75(6):3177-3198.
[34]徐越,李盈慧,宋懷波,等.基于Snake模型與角點(diǎn)檢測(cè)的雙果重疊蘋果目標(biāo)分割方法[J].農(nóng)業(yè)工程學(xué)報(bào),2015,31(1):196-203.
XU Yue, LI Yinghui, SONG Huaibo, et al. Segmentation method of overlapped double apples based on snake model and corner detectors[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015,31(1):196-203.
[35]XIANG R, JIANG H, YING Y. Recognition of clustered tomatoes based on binocular stereo vision[J].Computers and Electronics in Agriculture, 2014,106(5): 75-90.
[36]盧軍,桑農(nóng).變化光照下樹(shù)上柑橘目標(biāo)檢測(cè)與遮擋輪廓恢復(fù)技術(shù)[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2014,45(4):76-81.
LU Jun, SANG Nong. Detection of citrus fruits within tree canopy and recovery of occlusion contour in variable illumination[J]. Transactions of the Chinese Society for Agricultural Machinery, 2014,45(4):76-81.
[37]BULANON D, KATAOKA T, OKAMOTO H, et al. Determining the 3-D location of the apple fruit during harvest[C] //Automation Technology for Off-Road Equipment Proceedings of the 2004 Conference.[S.l.]:[s.n.], 2004: 91- 97.
[38]李斌,Ning Wang,汪懋華, 等.基于單目視覺(jué)的田間菠蘿果實(shí)識(shí)別[J].農(nóng)業(yè)工程學(xué)報(bào),2010,26(10): 345-349.
LI Bin, WANG Ning, WANG Maohua, et al. In-field pineapple recognition based on monocular vision[J]. Transactions of the Chinese Society of Agricultural Engineering, 2010,26(10): 345-349.
[39]熊俊濤,鄒湘軍,彭紅星,等.擾動(dòng)柑橘采摘的實(shí)時(shí)識(shí)別與采摘點(diǎn)確定技術(shù)[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2014,45(8):38-43.
XIONG Juntao, ZOU Xiangjun, PENG Hongxing, et al. Real-time identification and picking point localization of disturbance citrus picking[J]. Transactions of the Chinese Society for Agricultural Machinery, 2014,45(8):38-43.
[40]楊慶華,劉燦,荀一, 等.葡萄套袋機(jī)器人目標(biāo)識(shí)別方法[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2013,44(8):234-239.
YANG Qinghua, LIU Can, XUN Yi, et al. Target recognition for grape bagging robot[J]. Transactions of the Chinese Society for Agricultural Machinery, 2013,44(8):234-239.
[41]羅陸鋒,鄒湘軍,熊俊濤,等.自然環(huán)境下葡萄采摘機(jī)器人采摘點(diǎn)的自動(dòng)定位[J].農(nóng)業(yè)工程學(xué)報(bào), 2015, 31(2): 14-21.
LUO Lufeng, ZOU Xiangjun, XIONG Juntao, et al. Automatic positioning for picking point of grape picking robot in natural environment[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(2): 14-21.
[42]INKYU S, LEHNERT C, ENGLISH A, et al. Peduncle detection of sweet pepper for autonomous crop harvesting-combined color and 3-D information[J]. IEEE Robotics & Automation Letters, 2017,2(2) :765-772.
[43]羅陸鋒, 鄒湘軍, 葉敏, 等. 基于雙目立體視覺(jué)的葡萄采摘防碰空間包圍體求解與定位[J]. 農(nóng)業(yè)工程學(xué)報(bào), 2016, 32(8): 41-47.
LUO Lufeng, ZOU Xiangjun, YE Min, et al. Calculation and localization of bounding volume of grape for undamaged fruit picking based on binocular stereo vision[J]. Transactions of the Chinese Society of Agricultural Engineering, 2016, 32(8): 41-47.
[44]BAC C W, ROORDA T, RESHEF R, et al. Analysis of a motion planning problem for sweet-pepper harvesting in a dense obstacle environment[J]. Biosystems Engineering,2015, 146(6):85-97.
[45]梁喜鳳, 王永維, 苗香雯. 番茄收獲機(jī)械手避障運(yùn)動(dòng)規(guī)劃[J]. 農(nóng)業(yè)機(jī)械學(xué)報(bào), 2008, 39(11): 89-93.
LIANG Xifeng, WANG Yongwei, MIAO Xiangwen. Motion planning of a tomato harvesting manipulator for obstacle avoidance[J]. Transactions of the Chinese Society for Agricultural Machinery, 2008, 39(11): 89-93.
[46]尹建軍, 武傳宇, Yang Simon X, 等. 番茄采摘機(jī)器人機(jī)械臂避障路徑規(guī)劃[J]. 農(nóng)業(yè)機(jī)械學(xué)報(bào), 2012, 43(12): 171-175.
YIN Jianjun, WU Chuanyu, YANG Simon X, et al. Obstacle-avoidance path planning of robot arm for tomato-picking robot[J]. Transactions of the Chinese Society for Agricultural Machinery, 2012, 43(12): 171-175.
[47]王冰心,王孫安,于德弘.基于選擇性注意機(jī)制的果實(shí)簇識(shí)別與采摘順序規(guī)劃[J].農(nóng)業(yè)機(jī)械學(xué)報(bào),2016,47(11):1-7.
WANG Bingxin, WANG Sunan, YU Dehong. Fruit cluster recognition and picking sequence planning based on selective attention[J]. Transactions of the Chinese Society for Agricultural Machinery, 2016,47(11):1-7.
[48]羅陸鋒, 鄒湘軍, 程堂燦,等. 采摘機(jī)器人視覺(jué)定位及行為控制的硬件在環(huán)虛擬試驗(yàn)系統(tǒng)設(shè)計(jì)[J]. 農(nóng)業(yè)工程學(xué)報(bào), 2017, 33(4):39-46.
LUO Lujin, ZOU Xiangjun, CHENG Tangcan, et al. Design of virtual test system based on hardware-in-loop for picking robot vision localization and behavior control[J]. Transactions of the Chinese Society of Agricultural Engineering, 2017, 33(4):39-46.
[49]BLANES C, ORTIZ C, MELLADO M. Assessment of eggplant firmness with accelerometers on a pneumatic robot gripper[J]. Computers and Electronics in Agriculture, 2015, 113(4): 44-50.
[50]陳燕, 蔡偉亮, 向和平, 等. 荔枝采摘器優(yōu)化設(shè)計(jì)與仿真[J]. 農(nóng)機(jī)化研究, 2013(8): 84-87.
CHEN Yan, CAI Weiliang, XIANG Heping, et al. Optimized design and simulation of litchi harvesting device[J]. Journal of Agricultural Mechanization Research, 2013(8): 84-87.
[51]李娜, 王家忠, 張建寶, 等. 一種剛?cè)峄炻?lián)欠驅(qū)動(dòng)草莓采摘機(jī)械手設(shè)計(jì)[J]. 河北農(nóng)業(yè)大學(xué)學(xué)報(bào), 2015, 38(1): 118-121.
LI Na, WANG Jiazhong, ZHANG Jianbao, et al. Design of under actuated strawberry harvesting manipulator with rigid-compliant structured[J]. Journal of Agricultural University of Hebei, 2015, 38(1): 118-121.
[52]饒洪輝, 羅時(shí)挺, 徐諧慶,等. 一種氣吸式油茶果采摘機(jī)[P].中國(guó)專利: CN201520070541.1, 2015-02-06.
[53]王學(xué)林,姬長(zhǎng)英,周俊,等.基于灰色預(yù)測(cè)控制的果蔬抓取系統(tǒng)設(shè)計(jì)與試驗(yàn)[J].農(nóng)業(yè)工程學(xué)報(bào),2010, 26(3): 112-117.
WANG Xuelin, JI Changying, ZHOU Jun, et al. Design and experiment of fruit and vegetable grasping system based on grey prediction control[J]. Transactions of the Chinese Society of Agricultural Engineering, 2010, 26(3): 112-117.
[54]李建偉,陳艷艷.蘋果采摘機(jī)器人末端執(zhí)行器的原理及試驗(yàn)研究[J].農(nóng)機(jī)化研究, 2017(9):139-142.
LI Jianwei, CHEN Yanyan. Structural principle and experimental research on the end effector of apple picking robot[J]. Journal of Agricultural Mechanization Research, 2017(9):139-142.
[55]羅陸鋒, 楊自尚,徐東升,等.串型水果采摘機(jī)器人的末端執(zhí)行器[P]. 中國(guó)專利: CN201610843191.7, 2016-12-28.
[56]ZOU X, ZOU H, LU J. Virtual manipulator-based binocular stereo vision positioning system and errors modeling[J]. Machine Vision and Applications, 2012, 23(1): 43-63.
[57]LUO Lufeng, TANG Yunchao, ZOU Xiangjun, et al. Vision-based extraction of spatial information in grape cluster for harvesting robots[J]. Biosystems Engineering, 2016, 151(11):90-104.