【作者】 王玉香；

【导师】 田彦涛；

【作者基本信息】 吉林大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 机器视觉是随着60年代末计算机与电子技术的迅速发展而出现的。把视觉信息用于机械手定位的研究可以追溯到70年代，当时出现了一些实用性的视觉系统，如：应用于集成电路生产、精密电子产品装配等。80年代后期，出现了专门的图像处理硬件，使得视觉信息可用于连续的反馈，于是提出了基于视觉伺服的控制形式，即视觉伺服。这种方式可以克服模型中存在的不确定性，提高视觉定位或跟踪的精度。到了90年代，随着计算机能力的增强和价格下降，以及图像处理硬件和CCD摄像机的迅速发展，视觉伺服系统吸引了许多研究人员的注意。在过去的几年里，视觉伺服控制无论在理论还是在应用方面都有很大的进步。无约束运动体的轨迹控制问题正是以视觉伺服控制理论为基础而展开研究的，在实际应用中都具有极其重要的意义。本系统又称为板球系统。目前，在国外的一些大学中已经作为实验装置，用来学习和研究各种控制方法。搭建这个系统的目的主要是研究该系统的动力学问题，并将多种模糊控制等先进的控制理论运用到系统中，可以作为控制理论研究与开发的综合实验平台。 本文首先针对基于位置的视觉伺服系统的控制原理，设计出了实验平台的基本结构，并根据本系统具体的平台使用情况，从控制精度、性能以及价格等几个方面因素出发，适当的选择了平台各个部分所使用的元器件，并着重介绍了伺服控制器的工作原理及其应用。系统首先要对物体进行定位，在这里使用CCD摄像机作为视觉传<WP=83>感器，这就涉及到摄像机的标定问题。本系统由于采用的CCD摄像机不存在径向畸变，因此在线性摄像机模型的基础上，完成了摄像机的标定。然后将图像采集卡采集来的图像进行点运算，数学形态学处理，图像边缘检测和图像特征提取，将被控的小球从图像信息中提取出来，再与空间运动学及摄像机标定结果相结合，设计出摄像机固定在底座上，不随平台一起旋转情况下的小球的定位算法，并通过实验证明了定位算法的正确性。但是，系统定位算法的精度完全依赖于摄像机的标定精度，因此，不利于系统定位精度的提高。视觉控制器的设计从无约束运动物体的运动学、动力学分析入手，分析得出平台沿x,y两个方向能够完全解耦的结论。因此，进行模糊控制器的设计时x方向和y方向的设计是相同的。基本模糊控制器具有响应时间短，超调量小，鲁捧性好等优点。但是存在着稳态误差，并且对于具有复杂被控对象并且在不同的工作环境和控制要求下，却不能保证在系统的全部运行过程中都具有优良的控制性能，因此提出了进一步改善系统的性能的两个方法，即：位置预测控制和参数在线自整定模糊控制，并且对系统线后使用两种方法改善其控制性能。对于改善系统性能各种方法的有效性也通过仿真实验给与验证。本系统的任务给定部分，属于轨迹规划问题。轨迹规划属于机器人低层规划问题。轨迹规划是在物体运动学和动力学的基础上，研究运动体的轨迹生成方法的。本文从直角坐标空间中基本的点到点的规划方法人手，逐步深入到曲线的规划问题和过任意路径点的轨迹规划问题。设计了轨迹规划的插值方法，并通过实验及其仿真结果证实了轨迹规划的实时生成方法能够大幅度的减少系统的运行时间，并且能够使运动体在运行的过程中保持其运行速度平滑改变，达到运行平稳的目的。由于板球系统实验平台所具有的广泛实际意义，我们相信，本文所做的实验和理论、仿真的结果对以后的工作是有所贡献的。更多还原

【Abstract】 At the end of the 60s, machine vision appeared with the develop of the computer and electronic technique. Vision information has been used to orientate manipulator from 70s.At that time, there were some practical vision system, such as production of integrate circuit, assemblage of rigid electronic product. In the evening of the 80s, that there had appeared special hardware for image manipulation made visual information use to consecutive feedback. So, the control form based on visual servoing ,that is, visual servoing. This mode can conquer the uncertainty in the mode and improve the precision of the visual orientation or scout. At the 90s, with the improving of computer ability and the drop of price and the rapidly develop of hardware for image manipulation and CCD vidicon, visual servoing system attracted the attention of many people. In the last years, visual seroing control had many progress in the aspect both of theory and of applying.The issue of tracking control of unrestricted mobile is studied based on the theory of visual servoing. It has important meaning in the actual applying. This system can be also named as cricket system. Now, it has been used to study various control method as experimental device in some overseas university. The aim of building this system is to study the <WP=85>dynamical problems of it. Using the method of fuzzy control to cricket system can be used as a demo device for control lab of university and also exploiting flat for control theod.At first, this paper designed the basic framework of experimental flat based on the control theory of visual servoing system and properly chosen the organ used in the flat based on material use of flat from some aspects of control precision and capability and price and introduced the work theory of servo controller and its application.The system orientates the objects at first and uses the CCD vidicon as vision sensor so that it deals with the demarcating problem of the vidicon. This system has no radial aberrance because it adopts CCD vidicon. So, it can complete demarcation of the vidicon based the linear mode of the vidicon. Then it does point to point operation and morphological operations and edge detection and features extraction for the image from the image grabbing card so that it can distill the controlled globule from the image information, then combine it with the result of space kinematics and vidicon. The designed vidicon fixed on the bottom and we prove the correctness of the orientation algorithm. But, the accuracy of the orientation algorithm completely depends on the orientation accuracy. So, it is not propitious to improve orientation accuracy of the system.The design of the vision controller starts with the kinetics and dynamics analysis of unrestricted mobile which we can conclude that the flat can decouple along the x,y directions. So, the design in x direction is similar to the design in y direction. The basic fuzzy controller has the advantage of short responding time and small super adjust and good robust. But, there is stabilizing error and it can not have fine controlling capability in the entire run process of the system. Therefore, we advanced two means to improve the capability of the system which are position estimating control and parameter on-line self-settle fuzzy control and use orderly to the system to improve its control capability. We also validate the method improving the <WP=86>capability of the system.The task-given section belongs to path planning problem. Path planning belongs to the bottom planning problem. It studies the path building of the mobile based on kinematics and dynamics. This paper start with the basic point to point planning in the right angle space and gradually go deep into the path planning problem of curve planning and arbitrary path point. We design interpolation method of path planning and approve that the real time building method can reduce the run time of the system through the experiment and can make the mobile keep stable speed in the running process to realize the a更多还原