苹果套袋机器人视觉伺服技术研究

发布时间：2018-12-13 19:46

【摘要】：农业机器人研究的主要任务是将机器人工程学应用在农业上,或是根据作物特性探索利用现有的适用的工业机器人的可能性。机器人视觉伺服控制涉及较为广泛的研究内容,例如图像处理、特征提取、双目立体视觉标定、视觉控制算法、视觉测量等。本文以图像识别与视觉定位在苹果套袋系统中的应用为背景,以提高识别与定位系统的实时性和普适性为切入点,探索鲁棒性和实时性相融合的幼果图像识别与定位关键技术研究与方法,使之满足于实时应用的需要。主要研究内容如下:1、对苹果套袋机器人视觉系统进行总体介绍并给出具体流程图,对预处理的苹果幼果进行图像分割,运用改进后的连通域标记,面积法去除小面积以及圆形度等方法,对苹果幼果进行分割后的识别;采用双目立体视觉对左右摄像机分别采集并对最终识别出的苹果幼果图像进行立体匹配、图像极线校正以及三维重建,最后利用自主开发的苹果幼果上位机处理软件与OpenGL相结合对三维重建出的苹果幼果图像进行三维显示。2、对苹果幼果套袋机器人进行运动学建模,利用D-H方法对其建立运动学方程,对其进行正、逆运动学分析。利用基于Matlab的机械臂仿真工具箱Robotics Toolbox直观的显示出机械臂的可达工作空间以及参数的合理性。通过对机械臂的各个关节轨迹进行分析,为后续进行实际的套袋工作奠定良好的基础。3、分别介绍了基于图像的视觉伺服控制、基于位置的视觉伺服控制以及基于位置与基于图像的混合的2.5D视觉伺服控制,并分析了各自系统的特点以及稳定性。根据苹果套袋机器人实际的工作情况,本文选择基于位置与基于图像的混合视觉伺服控制作为的伺服控制系统。最后,在实验室条件下,模拟室外实际工作场景,对苹果幼果套袋机器人进行整体的系统联调,实验结果证明了方案的可行性和相关算法的有效性。
[Abstract]:The main task of agricultural robot research is to apply robotics engineering to agriculture, or to explore the possibility of utilizing existing applicable industrial robots according to crop characteristics. Robot visual servo control involves a wide range of research topics, such as image processing, feature extraction, binocular stereo vision calibration, visual control algorithm, visual measurement, and so on. In this paper, the application of image recognition and visual positioning in apple bagging system is taken as the background, and the real time and universality of the recognition and positioning system are improved as the breakthrough point. In order to satisfy the need of real-time application, the key techniques and methods of image recognition and localization of young fruit are explored. The main research contents are as follows: 1. The vision system of apple bagging robot is introduced and the flow chart is given. The small area and roundness were removed by the area method to recognize the young apple fruit after segmentation. The binocular stereo vision was used to collect the left and right cameras and to carry out stereo matching, polar line correction and 3D reconstruction of the final recognized apple fruit images. Finally, using the self-developed apple fruit processing software and OpenGL to display the 3D reconstructed apple fruit image. 2, the kinematics model of apple fruit bagging robot is established. The kinematics equation is established by D-H method, and the forward and inverse kinematics are analyzed. The reachable workspace and the reasonableness of the parameters of the manipulator are shown intuitively by using the arm simulation toolbox Robotics Toolbox based on Matlab. Through the analysis of each joint locus of the manipulator, it lays a good foundation for the practical bagging work. 3. The visual servo control based on the image is introduced respectively. The position based visual servo control and the 2.5D visual servo control based on the combination of position and image are presented. The characteristics and stability of the respective systems are analyzed. According to the actual working situation of Apple bagged robot, this paper chooses the hybrid visual servo control based on position and image as the servo control system. Finally, under the laboratory condition, simulating the outdoor actual working scene, the whole system of the apple fruit bagging robot was adjusted. The experimental results proved the feasibility of the scheme and the effectiveness of the related algorithm.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP242

【参考文献】