基于RGB-D传感器的新型弧焊机器人示教系统

CORC > 自动化研究所 > 中国科学院自动化研究所 > 毕业生 > 硕士学位论文

题名	基于RGB-D传感器的新型弧焊机器人示教系统
作者	李婧
学位类别	工学硕士
答辩日期	2016-05-24
授予单位	中国科学院大学
授予地点	北京
导师	景奉水
关键词	焊接机器人示教器点云机器人操作系统（ROS）路径点生成
学位专业	控制理论与控制工程
中文摘要	工业机器人经过几十年来的长足发展，在控制精度上已经达到了很高的要求。但在工厂中广泛使用的工业机器人还是需要人类操作员来进行耗时的示教工作。工业机器人大多缺乏对环境的感知能力，只能按部就班工作，距离真正智能还有一段差距。本文针对工业机器人焊接作业，提出了一种新的机器人示教系统解决方案，旨在减轻工人示教操作的工作负担:在工业机器人末端增加装置并安装RGB-D深度相机，通过深度相机获取工作环境的信息，用上位机读取点云并生成路径点并下发控制器和示教器，操作员可以示教器编辑这些路径点并生成示教程序。论文的主要内容如下： 1）总结了工业机器人的发展情况以及示教器的研究现状。对于示教过程中的数学模型和传感器原理进行了阐述，并提出了新型示教系统的整体方案。 2）提出了从点云中提取路径点的算法。包括从三维点云中分离出二维图像，进行ROI设定、滤波，焊缝细节加强、焊缝边缘提取等图像处理步骤，并结合三维点云中深度信息获取路径点的坐标和对应的焊枪位姿。 3）提出了用于焊接机器人的示教器软件架构并进行了软件开发。针对示教器与上位机协同作业的需求，增加了示教器对上位机的通讯接口。 4）基于ROS平台，实现了从点云中生成路径点的算法，并进行了实验验证。最后对全文工作进行了总结，并对下一步工作进行了展望。
英文摘要	After decades of rapid development, industrial robots have realized hi-precision servo control. But the most widely used industrial robots in the factory still need human operator using a teaching pendant to program the robot, which is a time-consuming and tired work. Most industrial robots are lack of perception of the environment, cannot decide what to do, still far from being called intelligent. A new robot teaching system aiming at industrial robot’s arc welding task is developed in this paper, hoping to ease the workload of robot programmer. By installing an RGB-D depth camera at the end of robot, an upper computer can obtain the point cloud of work environment. This upper computer is used to extract the path points, which is sent it to controller and teach pendant, from point cloud. The operators may modify these path points by teach pendant and generate a robot program. The main content of this paper as follow: 1) This paper summarized the development situation of industrial robot and research status of teach pendant. We analyzed the mathematic model of robot programming and basic principle of the sensor, and proposed a new teaching system solution. 2) The algorithm of extracting path points from point cloud is proposed. Including procedures like separating 2D image in 3D point cloud, setting ROI and filter, extracting the edge of welding line and find path points after intensifying the detail of the welding line and other image process, obtaining the 3D coordinate from 3D point cloud and calculate the pose of path points and so on. 3) This paper proposed the architecture of the software and developed the software. Aiming to the need of teach pendant communicating with upper computer, a communication interface to upper computer is added. 4) Conducted experiments which are implemented with Python in Robot Operating System (ROS) to verify the accuracy of the algorithm which generate path points from point cloud. Finally, summarized the full text and gives the prospect for further work.
语种	中文
内容类型	学位论文
源URL	[http://ir.ia.ac.cn/handle/173211/11769]
专题	毕业生_硕士学位论文
作者单位	中国科学院自动化研究所
推荐引用方式 GB/T 7714	李婧. 基于RGB-D传感器的新型弧焊机器人示教系统[D]. 北京. 中国科学院大学. 2016.