采摘机器人微扰目标与视觉定位的耦合机理

项目名称： 采摘机器人微扰目标与视觉定位的耦合机理

项目编号： No.31201135

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 食品科学、农学基础与作物学

项目作者： 熊俊涛

作者单位： 华南农业大学

项目金额： 21万元

中文摘要： 针对复杂环境下果蔬采摘机器人作业时，目标受到扰动难以精准定位的问题，研究采摘目标微扰动的受力与视觉定位的耦合机理。首先从力学角度定量分析目标扰动状态，研究扰动幅度与受力的关系，建立扰动力的关系模型，确定扰动状态的辨识参数，实现扰动状态的量化与分级；然后研究扰动力与视觉图像模糊程度的关系，进行模糊图像分类；进一步探索基于图像分类的模糊图像处理方法，通过改进图像复原算法中的离散余弦变换法，有效求取图像模糊算子的正则化逆，实现微扰动荔枝串边缘模糊的图像复原算法和果实的识别定位算法；最后对视觉定位行为进行系统综合分析，建立微扰动条件下视觉定位行为的解耦模型和智能决策知识库，实现视觉系统对目标扰动状态的主动辨识，并进行智能决策与定位计算。用自制的双目视觉采摘机器人物理样机和定量仿真方法,对视觉定位机制进行验证。本项目旨在探索微扰动目标的视觉定位机理，为机器人的视觉精准定位奠定理论和技术基础。

中文关键词： 采摘机器人；扰动状态；视觉定位；智能系统；

英文摘要： The coupling mechanism of the force and the visual position in picking target weak perturbation is researched for the problem of that the harvesting robots in complex environment are difficult to position precisely when disturbed.First the target perturbation state is analyzed quantitatively from mechanics angles, the relationship between the perturbation range and the force is researched to set up the relation model of the disturbing force, the identification parameters estimation are determined to realize the quantization and classification of the perturbation;then the relationship between the disturbing force and the visual image fuzzy degree is researched to realize the classification of fuzzy image. Further explore the fuzzy image processing methods based on image classification, through improving the discrete cosine transform method in the image restoration algorithm, effectively solve the regularized inverse of the image fuzzy operator, realize the image restoration algorithm of the edge blurring resulting from weak perturbation of litchi and its location algorithm; finally comprehensively analyze the visual position behavior, and the decoupling model and intelligent decision knowledge base in weak perturbation condition are built to realize the active identification for the target perturbation state of t

英文关键词： picking robots；disturbance state；visual positioning；intelligent system；