基于视觉引导的工业机器人应用研究

发布时间：2018-02-01 23:37

本文关键词： 视觉引导工业机器人目标识别目标定位分拣　出处：《陕西科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：智能制造已成为生产制造业的发展方向,其核心是工业机器人。在工业生产中应用工业机器人可以提高产品的产量与质量并降低生产成本。为工业机器人配备视觉引导系统,可以提高机器人对外界环境的感知和适应能力,对实现制造业的智能化有非常重要的意义。本文以分拣系统为背景,研究了视觉引导技术在工业机器人中的应用,具体的工作可分为以下三个部分:(1)基于轮廓匹配的目标识别和定位方法研究。首先利用基于边缘的方法将图像分割成相互独立的小块图像,然后提取图像中目标物体的轮廓特征,分别计算目标物体轮廓图像和模板轮廓图像的几何不变矩(Hu矩)和两个轮廓之间的对比度量值,最后根据度量值阈值去除错误的匹配结果。利用图像的标准矩计算出目标物体的中心像素坐标,分别利用目标图像和模板图像的中心坐标和重心坐标构成向量,计算两个向量之间的夹角,并根据实际系统对夹角做一定的偏移,得到目标物体相对于模板图像的旋转角度。(2)基于特征点匹配的目标识别和定位方法研究。首先利用基于边缘的方法将图像分割成相互独立的小块图像,然后检测目标物体和模板图像上的特征点,分别计算对应特征点之间的汉明距离,并根据阈值判断特征点是否匹配,最后根据最佳特征点的个数去除掉错误的匹配结果。分别利用目标图像和模板图像的中心坐标以及任意的两个特征点坐标构成的三角形,计算出目标物体的像素坐标,并利用任意两个特征点构成的特征点向量,计算两个向量之间的夹角,并做一定的偏移,得到目标物体相对于模板图像的旋转角度。(3)基于视觉引导的分拣系统设计。系统以简单工件和象棋棋子作为分拣对象,首先利用工业数字相机获取工作区域的图像,在工业控制计算机上利用C++编写视觉处理程序实现目标的识别和定位,并采用以太网通信将目标信息数据发送到工业机器人,然后利用MELFA-BASIC语言编写机器人程序实现数据的解析及对工业机器人的控制,最后采用三菱RV-13F六自由度工业机器人作为主体,气动吸嘴作为末端执行器抓取目标物体,实现自动分拣。为了验证目标识别算法的有效性,进行了目标识别与定位实验,与利用示教器获取的目标实测世界坐标进行比较,利用绝对误差分析识别定位结果的准确性。为了检验分拣系统的性能,进行了物体的抓取与分拣实验。实验结果表明,计算坐标与实测世界坐标之间的误差在0.6mm以内,可以准确的抓取到目标物体,同时利用计算出的旋转角度可以将图像区域中的所有目标物体按照模板图像的方向整齐摆放,实现了物体的自动分拣。
[Abstract]:Intelligent manufacturing has become the development direction of manufacturing industry. The application of industrial robots in industrial production can improve the output and quality of products and reduce the cost of production. The industrial robots are equipped with visual guidance system. It can improve the perception and adaptability of the robot to the outside environment, and it is very important to realize the intelligence of manufacturing industry. This paper takes the sorting system as the background. The application of visual guidance technology in industrial robot is studied. The specific work can be divided into the following three parts: 1) Target recognition and localization based on contour matching. Firstly, the edge based method is used to segment the image into independent blocks. Then the contour features of the target object in the image are extracted, and the geometric invariant moments Hu moments of the contour image and the template contour image are calculated, respectively, and the contrast measures between the two contours are calculated. Finally, according to the measure threshold to remove the wrong matching results, the center pixel coordinates of the target object are calculated by using the standard moments of the image, and the center coordinates and the barycentric coordinates of the target image and the template image are used to form the vector respectively. The angle between the two vectors is calculated, and a certain deviation of the angle is made according to the actual system. The rotation angle of the target object relative to the template image is obtained. (2) the method of target recognition and location based on feature point matching is studied. Firstly, the image is segmented into independent small images by edge based method. Then the feature points of the target object and the template image are detected, the hamming distance between the corresponding feature points is calculated, and the matching of the feature points is judged according to the threshold value. Finally, according to the optimal number of feature points to remove the wrong matching results, respectively using the target image and template image center coordinates and arbitrary two feature point coordinates of the triangle. The pixel coordinates of the target object are calculated, and the angle between the two vectors is calculated by using the eigenpoint vector of any two feature points, and a certain deviation is made. The rotation angle of the target object relative to the template image is obtained.) the visual guided sorting system is designed. The system takes the simple workpiece and the chess piece as the sorting object. Firstly, the image of the working area is obtained by using the industrial digital camera, and the visual processing program is written on the industrial control computer to realize the recognition and localization of the target. The target information data is sent to the industrial robot by Ethernet communication, and then the robot program is written by MELFA-BASIC language to realize the data analysis and the control of the industrial robot. Finally, Mitsubishi RV-13F six-degree-of-freedom industrial robot is used as the main body, and the pneumatic suction nozzle is used as the end actuator to grab the target object to realize automatic sorting, in order to verify the effectiveness of the target recognition algorithm. The experiments of target recognition and location are carried out, and compared with the measured world coordinates of the target obtained by the teacher, and the accuracy of the identification and location results is analyzed by using absolute error. In order to test the performance of the sorting system. The experimental results show that the error between the calculated coordinates and the measured world coordinates is within 0.6 mm, and the object can be captured accurately. At the same time, all the target objects in the image region can be arranged neatly according to the direction of the template image by using the calculated rotation angle, and the automatic sorting of objects is realized.
【学位授予单位】：陕西科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP242.2

【参考文献】