本文關(guān)鍵詞： 對靶噴霧 視覺伺服 軌跡規(guī)劃 機器人 手眼系統(tǒng)　出處：《中國農(nóng)業(yè)大學(xué)》2017年博士論文　論文類型：學(xué)位論文

【摘要】：對靶噴霧技術(shù)對減少農(nóng)藥使用量、降低作物農(nóng)藥殘留、保護環(huán)境和勞動者具有重要意義。本文針對溫室作物對靶噴霧作業(yè),提出了一種基于視覺伺服的噴霧機器人方案并構(gòu)建了試驗系統(tǒng),研究了基于該方案的視覺伺服對靶控制算法。主要研究內(nèi)容如下:(1)對靶噴霧機器人方案設(shè)計與試驗系統(tǒng)構(gòu)建。提出了一種基于單目場景/單目手眼混合視覺結(jié)構(gòu)的對靶噴霧機器人方案;根據(jù)該方案選定了噴霧機械臂的結(jié)構(gòu)形式,對噴霧機械臂進行了運動學(xué)位置正逆解、正逆向微分動力學(xué)及工作空間分析,并根據(jù)所需工作空間對其主要結(jié)構(gòu)參數(shù)進行了優(yōu)化,根據(jù)優(yōu)化結(jié)果確定了機械臂型號;設(shè)計了噴霧機械臂在笛卡爾空間和關(guān)節(jié)空間的軌跡插值方法;構(gòu)建了對靶噴霧機器人試驗系統(tǒng),完成了試驗系統(tǒng)的軟硬件設(shè)計。(2)作物圖像背景分割與靶標(biāo)信息提取方法研究。以定植初期的葫蘆幼苗為研究對象,通過對比各顏色空間常用算子對樣本圖像的背景分割效果,發(fā)現(xiàn)RGB空間的CIVE與2G-R-B算子的分割效果最好;以樣本圖像灰度直方圖的重疊區(qū)域占比、均值間距、標(biāo)準(zhǔn)偏差及算法耗時等為衡量指標(biāo),采用遺傳算法在RGB空間內(nèi)搜索,獲得了性能更優(yōu)的分割算子;鑒于該算子對作物占比較小的圖像分割效果不理想,提出了一種該算子+ Ostu分割法與a分量固定閾值分割法相結(jié)合的組合分割策略;針對選取的作物形心、包圍圓等對靶特征,設(shè)計了作物連通區(qū)域提取方法和包圍圓特征提取算法。(3)噴霧機器人視覺伺服控制研究。針對噴霧機器人選用的速度控制器,設(shè)計了一組基于作物冠層包圍圓特征計算的噴頭位置偏差和一組期望值不隨作業(yè)目標(biāo)更換而變化的矩特征等兩種反饋特征,室內(nèi)試驗證明,選用兩種反饋特征的控制器均具有較好的對靶及跟蹤控制性能。此外,針對噴霧機器人的連續(xù)運動作業(yè)模式,提出了一種基于改進勢場法的噴霧機械臂運動規(guī)劃對靶實時控制算法,該算法在傳統(tǒng)勢場法基礎(chǔ)上引入了速度勢場、視場約束和關(guān)節(jié)極限位置約束,可規(guī)劃出更優(yōu)的噴霧機械臂運動軌跡,有效提高噴霧機器人對靶控制的魯棒性。針對噴霧機器人的間歇運動作業(yè)模式,在確定各目標(biāo)作物噴霧點及各噴霧點兩兩間可行路徑的基礎(chǔ)上,采用枚舉法與改進近鄰法相結(jié)合的組合策略進行噴霧路徑規(guī)劃,并控制噴霧機械臂對靶,試驗驗證了該算法的可行性。
[Abstract]:Target spray technology to reduce the use of pesticides, crops reduce pesticide residues, has important significance to protect the environment and workers. According to the operation of greenhouse crops target spray, based on spraying robot visual servo scheme and the construction of the test system based on the scheme of visual servo control algorithm. The main research target the contents are as follows: (1) the construction of system design and test target spraying robot. This paper proposes a scheme based on monocular scene / single hand eye vision mixed structure on target spray robot scheme; according to the program selected the structure of mechanical arm to spray, spray manipulator kinematic position of forward and inverse solution analysis, reverse differential dynamics and the work space, and according to the required working space to optimize the main structure parameters, according to the optimization results of manipulator models were determined; the design of spray Trajectory interpolation method of fog manipulator in Cartesian space and joint space; construct the target spraying robot test system, designs the hardware and software of the test system. (2) crop image background segmentation and target information extraction method. In the early colonization of pumpkin seedlings as the research object, through the comparison of the commonly used color space operator based on the sample image segmentation effect, found that the best CIVE RGB space and 2G-R-B operator segmentation; gray histogram with overlapping sample image ratio, mean spacing, standard deviation and time-consuming algorithm as the measure, the genetic algorithm is used to search in the RGB space, the segmentation operator has better performances in view; the operator of the crops accounted for relatively small image segmentation effect is not ideal, put forward a kind of the operator + Ostu segmentation method and a component fixed threshold segmentation method combining the segmentation strategy; For the selected crops on the target centroid, surrounded by circular features, design method of connected region extraction and surrounded by the crop circle feature extraction algorithm. (3) study of visual servo control for speed controller. The robot spraying robot selection, a set of nozzle position deviation of crop canopy surrounded by circular features and a set of calculation based on the expected value do not change with changing job target moment features two kinds of feedback feature design, experiment show that the selection of two kinds of feedback characteristics of the controller have better targeting and tracking control performance. In addition, according to the operation mode of continuous motion of robot, proposes an improved potential field method of spray manipulator motion planning the target real-time control algorithm based on the algorithm in the method based on the introduction of the traditional potential field velocity potential field, field constraint and joint limit position constraint, can plan out the sprayer better Mechanical arm motion trajectory, effectively improve the robustness of the spraying robot target control. For intermittent operation mode of spraying robot, based on determining the target point and the crop spray spray point 22 feasible paths, combination strategies and improved nearest neighbor method combining the enumeration method of path planning and control of spray, spray the mechanical arm of the target, test and verify the feasibility of the algorithm.

【學(xué)位授予單位】：中國農(nóng)業(yè)大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2017
【分類號】：TP391.41;TP242

【參考文獻】

相關(guān)期刊論文 前10條

1 紀(jì)鵬;宋愛國;吳常鋮;魏宏明;宋子墨;張琪;;適用于移動機械手無關(guān)節(jié)狀態(tài)反饋情況的基于人-機-機協(xié)作的無標(biāo)定視覺伺服控制[J];機器人;2017年02期

2 吳作君;劉國華;;自適應(yīng)環(huán)境的機器人視覺伺服控制方法[J];計算機測量與控制;2016年12期

3 紀(jì)明山;;植保無人機應(yīng)用現(xiàn)狀與發(fā)展趨勢[J];新農(nóng)業(yè);2016年20期

4 孫瑛;程文韜;李公法;孔建益;蔣國璋;李U，

本文編號：1549656