全自动蔬菜嫁接机盘上匹配方法及自适应贴接法嫁接机构研究

Following the agronomic requirement of grafting operation, an on-disk scion-rootstock measurement and matching approach was proposed and auto-adaptive pith-cambium grafting mechanism for automatic grafting robot was simultaneous put forward in this proposal. Main research contents of this proposal are as follow: Study of plug seedlings characteristic parameters measurement approach under complicated aperture disk condition based on stereo vision technology. Study of parallel computing method for image processing with multiple cores and multiple threads. Study of the geometrical model for single plug seedling. Study the matching relationship between scions and rootstocks, and also the unmatching tolerance. Study of dynamic optimal matching approach of plug seedling sequences of scions and rootstocks. Study of optimal cut-position positioning method for both scion vegetative point and rootstocks hypocotyls. The development of a self-adaptive cutting and grafting mechanism under the driving of machine vision. System integration of vision measurement subsystem, seedlings loading subsystem, grafting mechanisms, ect.. The proposed research aimed at a further improvement of delicate operating performance of automatic fruit and vegetable grafting robot, so as to improve firm technical supports for grafting machinery industry, and dedicated in raising ston survival rate after grafting operation so as to increase the benefit of agriculture equipments and promote grafting planting, which do help to farming safety and increasing benefits of fruit and vegetable products, and lead to the successive development of economic agriculture in our country. The proposed methods such as plug seedlings characteristic parameters measurement under complicated aperture disk condition based on stereo vision technology and the construction of self-adaptive grafting mechanism under the driving of machine vision are also capable of other utilities.

根据嫁接作业农艺要求，本项目提出在穴盘上直接进行砧木和接穗生长特征参数检测和相互配对的盘上匹配方法研究，并开展具有穴苗个体情况自适应调制的嫁接机构研制。主要研究内容包括：穴盘复杂条件下基于三维立体视觉的穴苗生长状态特征参数检测方法研究，多核多线程并行处理方法的研究，穴苗模型研究，砧木-接穗胚径最佳匹配关系和失配容忍度研究，砧木-接穗动态优化匹配方法研究，穴苗生长点最佳切除位置和断根点定位方法研究，视觉驱动的自适应嫁接机构研制，视觉检测、自动上苗、嫁接机构等各子系统集成等。本项目研究提高全自动果蔬嫁接机作业的精确化程度，提升我国嫁接机行业的技术水平，提高砧穗成活率，推动嫁接种植从而提高设施农业效益，以及果蔬生产的安全性和效益，促进我国经济农业可持续发展。研究提出的穴盘复杂条件下穴苗生长状态特征参数检测方法、自适应嫁接机构等可为同类研究提供参考。

果蔬嫁接的意义在于解决连作障碍，机械化的嫁接作业是必由途径，因此我国积极开展了嫁接机研究，以满足设施栽培规模化生产模式的要求。本项目开展的工作和取得的结果主要包括以下几个方面：.多视觉的盘上穴盘苗参数检测方法研究。建立了穴盘苗图像微距采集系统，提出了苗茎和子叶图像处理方法；改进了张氏立体视觉系统标定方法，特征点检测误差降低9.6%，提取速度提高14.8%；实现基于激光-视觉的空间测量方法，94.5%次的苗径测量与人工测量差值小于0.1mm；利用立体视觉实现子叶轮廓三维坐标恢复，在高差突变处实现子叶分离；在四核嵌入式系统上实现穴盘苗图像并行处理，性能提高56%-65%。.穴盘苗模型及砧木-接穗匹配方法研究。提出了基于椭圆拟合的叶面恢复和参数提取方法，拟合成功率97.5%；穴盘苗模型研究表明：20天西葫芦苗茎短轴和长轴分别位于子叶展向的61.23º和151º处，20天西瓜苗的子叶展向与短轴方向相差在±5º范围内，为穴盘苗定向取苗提供基础；匹配嫁接关系研究表明：3.15-3.5mm苗径的葫芦苗砧木与2.5-2.8mm苗径西瓜接穗、3.5-3.85mm苗径的葫芦苗砧木与2.5-2.8mm苗径西瓜接穗具有最高的嫁接成功率与成活率，为匹配嫁接提供依据。.基于视觉驱动的自适应嫁接机构研究方面。切削试验研究表明：葫芦苗茎短轴和长轴方向最大剪切力分别为0.198N和0.207N，西瓜苗茎短轴和长轴方向最大剪切力分别为0.085N和0.091N，葫芦苗砧木子叶的最大容许切削角度偏差为±25º，对剪切机构体的设计有指导意义；提出了基于形态特征的嫁接抓取点定位方法和砧木生长点机器视觉定位方法，正确识别无腋芽情况，对有腋芽的识别正确率91%；开展了夹持、调整、上苗等机构的设计，可动态调整上苗高度、方向和切削角度等参数，取得9项专利。.嫁接试验机分析与构建。设计了二自由度穴盘苗取苗装置的研究设计，综合取苗速度6.66棵/分钟，嫁接成功率为87%；开展回栽参数优化试验、终端执行器研制及回栽试验，当基质孔径8mm、深度25mm，镇压块为半径20mm、端面斜度5°的半圆柱体时，取得92%的回栽成功率，对回栽机构的设计有指导意义。