【作者】 陈利兵；

【导师】 张铁中；

【作者基本信息】 中国农业大学 ， 机械电子工程， 2005， 硕士

【摘要】 为了实现草莓收获的自动化以及智能化,在前人研究的基础上,对草莓收获机器人采摘系统进行了研究。搭建了一个三自由度的直角坐标机械手,设计了该机器人的视觉系统,并为机器人建立了一个开放式的控制系统,最后重点对草莓图像处理算法以进行了研究,并编程实现。 开放式机器人控制器可以提高系统的柔性、可扩展性、可靠性和复用性等性能。以工控机和GALIL运动控制器为核心,建立模块化、开放式的机器人控制系统。采用VC++语言来编写系统程序,通过矢量运动节省作业时间。试验结果表明,控制系统运行良好。 采用LRCD(luminance and red color difference)方法分割草莓图像。在RGB色彩模型中,求得图像中每个像素的色差,在灰度图像中显示以色差值为灰度值的色差图像,取合适的阈值对该图像二值化,得到分割后的草莓图像;提取分割后草莓图像的几何特征,由边界到重心的最长距离来确定草莓果尖的位置,并在过果尖与重心的反相延长线上取一合适的点(例如,本研究中取其到重心的距离与果尖到重心的距离相等)作为采摘点。并通过坐标变换,获得该采摘点在空间上的实际坐标值。试验结果表明,视觉系统对采摘点位置的判断误差<3mm。 该开放式机器人控制系统能够准确到达指定位置,其图像处理方法切实可行。更多还原

【Abstract】 To automatize and intelligentize the process of strawberry-collecting, the picking system of strawberry-harvesting robots is studied on the previous basis. A Cartesian coordinate manipulator with three degree-of-freedoms (DOFs) and an open structure control system are set up, at the same time the vision system is designed, then the arithmetic of image process is especially researched and programmed.the controller has open structure so as to enhance the flexility, the extension, the reliability and the repetitious use of the system. With the core of PC and GALIL Controller, the open and modular control system is set up. The system program is written by VC++. The vector movement can save the working time. The result shows that the system operates wellLRCD(luminance and red color difference) method is adopted to segment image of strawberry. The result are shown in a gray image. Segmented image is gained by translating the gray image to a binary image after selecting an appropriate threshold. After extracting geometry feature of segmented strawberry image the tip position can be ascertained by the longest distance from the boundary to the barycenter, then the plucking position is gained in the reverse line from the tip to the barycenter(the distance from the plucking position to the barycenter is equal to that from the tip to the barycenter). The plucking position’s actual coordinates can be got by transforming the coordinates. The result reveals that the eccentricity of the plucking position will be no more than 3 mm in the robot vision system.This open control system can arrive the position specified well and truly and the method to process images is feasible.更多还原