纳米操作机视觉伺服技术研究
发布时间：2018-11-18 07:29
【摘要】：纳米技术在飞速发展,纳米材料依靠其优良独特的性能应用于各个领域。纳米操作技术对纳米科技的发展起到至关重要的作用。以往的纳米操作系统都是依靠人手工操作信号输入设备完成对纳米器件的操作,这种方法很费时,还容易造成设备损坏。随着机器人技术的快速发展,实现纳米操作的自动化是大势所趋。纳米操作的自动化必须实现闭环控制,然而现有的传感器很难达到纳米级别的测量精度,扫描电子显微镜(SEM)凭借其极高的分辨率、高采样率成为纳米操作系统发展必不可少的工具。本文提出依靠扫描电镜图像,利用视觉伺服技术实现一定程度上自动化的纳米操作机,总结了用于纳米操作机的视觉伺服技术并对这些方法在纳米操作机上的应用效果进行了仿真测试。首先搭建了纳米操作系统,分析了各组成部分的作用。选择SEM作为视觉伺服系统图像信息采集反馈装置,简要介绍了扫描电镜的成像原理和系统选用电镜的型号和性能指标;选择压电陶瓷驱动器作为探针控制器件,得到压电陶瓷器件的简化模型;同时为了实现系统的快速性,在位置反馈环节依靠扫描电镜和应变片共同完成。由于整个系统性能很大程度上依赖于SEM图像,需要对图像进行高性能处理,包括降噪和图像漂移补偿。选择NL-means算法作为去噪滤波器,相对于传统的降噪方法有更优良的性能;但这种方法运算量大,普通的处理方式不能满足伺服系统的实时性要求,随后采用依靠GPU并行加速的方法解决去噪的快速性问题;提出了将图像相似性算法用于对SEM图像漂移补偿的方案;作为实现视觉伺服的必备条件,必须知道被控对象和目标的实时位置,本文利用具有仿射不变性的SIFT算法实现对探针针尖的运动,对于追踪目标的辨识,依据具体使用条件做出调整。课题的目标是通过SEM图像获得压电驱动器所需的输入信号,由图像信息作为控制规律的输入信号得到。将小孔成像模型的雅可比模型迁移到SEM成像模型上,依靠在线调节算法获得实时图像雅可比矩阵;针对探针控制快速性和精确性的综合要求,实现基于应变片信号和基于视觉伺服的信号切换方法;提出一种简单而实用性强的Z轴接触检测方法。最后,在图像平面上进行了目标跟踪仿真,并对仿真结果做出分析,从而得到最优的控制参数;利用VRML软件搭建起三维笛卡尔坐标系下的视觉伺服控制仿真平台,模拟了视觉伺服的实现过程。
[Abstract]:Nanotechnology is developing rapidly. Nanomaterials are applied in various fields depending on their excellent and unique properties. Nanotechnology plays an important role in the development of nanotechnology. In the past, the nanoscale operating system used to rely on human manual operation signal input equipment to complete the operation of nano-devices, this method is time-consuming, but also easy to cause damage to the device. With the rapid development of robot technology, it is the trend of the times to realize the automation of nano-operation. The automation of nanoscale operation must be controlled by closed loop. However, it is very difficult for the existing sensors to achieve the measurement accuracy of nanometer level. The scanning electron microscope (SEM) has its very high resolution. High sampling rate has become an essential tool for the development of nanoscale operating systems. In this paper, we propose to realize a certain degree of automation of nanoscale manipulators by means of scanning electron microscope images and visual servo technology. In this paper, the visual servo technology used in nano manipulators is summarized, and the application effect of these methods in nano manipulators is tested by simulation. First, the nanometer operating system is built, and the function of each component is analyzed. SEM is selected as the image information collection and feedback device of the visual servo system. The imaging principle of the scanning electron microscope and the model and performance index of the system are briefly introduced. The piezoelectric actuator is selected as the probe controller to obtain the simplified model of the piezoelectric ceramic device, and in order to realize the rapidity of the system, the scanning electron microscope and strain gauge are used to achieve the position feedback. Since the performance of the whole system depends heavily on SEM images, it is necessary to perform high performance image processing, including noise reduction and image drift compensation. The NL-means algorithm is chosen as the denoising filter, which has better performance than the traditional de-noising method. However, this method can not meet the real-time requirement of servo system because of its large computation, and then it adopts the method of parallel acceleration of GPU to solve the problem of fast de-noising. An image similarity algorithm is proposed to compensate SEM image drift. As a necessary condition to realize visual servo, we must know the real time position of the controlled object and the target. In this paper, we use the affine invariant SIFT algorithm to realize the motion of the probe tip and identify the tracking target. Make adjustment according to specific use condition. The object of this paper is to obtain the input signal of piezoelectric actuator by SEM image, which is obtained by image information as the input signal of control law. The Jacobian model of the pinhole imaging model is migrated to the SEM imaging model, and the real time image Jacobian matrix is obtained by on-line adjusting algorithm. Aiming at the comprehensive requirement of the speed and accuracy of probe control, the method of signal switching based on strain gauge signal and visual servo is realized, and a simple and practical Z-axis contact detection method is proposed. Finally, the target tracking simulation is carried out on the image plane, and the simulation results are analyzed to obtain the optimal control parameters. The visual servo control simulation platform in 3D Cartesian coordinate system is built by using VRML software, and the realization process of visual servo is simulated.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TB383.1;TP391.41
本文编号：2339319
[Abstract]:Nanotechnology is developing rapidly. Nanomaterials are applied in various fields depending on their excellent and unique properties. Nanotechnology plays an important role in the development of nanotechnology. In the past, the nanoscale operating system used to rely on human manual operation signal input equipment to complete the operation of nano-devices, this method is time-consuming, but also easy to cause damage to the device. With the rapid development of robot technology, it is the trend of the times to realize the automation of nano-operation. The automation of nanoscale operation must be controlled by closed loop. However, it is very difficult for the existing sensors to achieve the measurement accuracy of nanometer level. The scanning electron microscope (SEM) has its very high resolution. High sampling rate has become an essential tool for the development of nanoscale operating systems. In this paper, we propose to realize a certain degree of automation of nanoscale manipulators by means of scanning electron microscope images and visual servo technology. In this paper, the visual servo technology used in nano manipulators is summarized, and the application effect of these methods in nano manipulators is tested by simulation. First, the nanometer operating system is built, and the function of each component is analyzed. SEM is selected as the image information collection and feedback device of the visual servo system. The imaging principle of the scanning electron microscope and the model and performance index of the system are briefly introduced. The piezoelectric actuator is selected as the probe controller to obtain the simplified model of the piezoelectric ceramic device, and in order to realize the rapidity of the system, the scanning electron microscope and strain gauge are used to achieve the position feedback. Since the performance of the whole system depends heavily on SEM images, it is necessary to perform high performance image processing, including noise reduction and image drift compensation. The NL-means algorithm is chosen as the denoising filter, which has better performance than the traditional de-noising method. However, this method can not meet the real-time requirement of servo system because of its large computation, and then it adopts the method of parallel acceleration of GPU to solve the problem of fast de-noising. An image similarity algorithm is proposed to compensate SEM image drift. As a necessary condition to realize visual servo, we must know the real time position of the controlled object and the target. In this paper, we use the affine invariant SIFT algorithm to realize the motion of the probe tip and identify the tracking target. Make adjustment according to specific use condition. The object of this paper is to obtain the input signal of piezoelectric actuator by SEM image, which is obtained by image information as the input signal of control law. The Jacobian model of the pinhole imaging model is migrated to the SEM imaging model, and the real time image Jacobian matrix is obtained by on-line adjusting algorithm. Aiming at the comprehensive requirement of the speed and accuracy of probe control, the method of signal switching based on strain gauge signal and visual servo is realized, and a simple and practical Z-axis contact detection method is proposed. Finally, the target tracking simulation is carried out on the image plane, and the simulation results are analyzed to obtain the optimal control parameters. The visual servo control simulation platform in 3D Cartesian coordinate system is built by using VRML software, and the realization process of visual servo is simulated.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TB383.1;TP391.41
【参考文献】
相关期刊论文 前9条
1 李东洁;王金玉;尤波;肖万哲;;纳米操作研究现状及发展趋势[J];哈尔滨理工大学学报;2011年06期
2 王述彦;师宇;冯忠绪;;基于模糊PID控制器的控制方法研究[J];机械科学与技术;2011年01期
3 田孝军;王越超;席宁;董再励;;基于AFM的机器人化纳米操作中纳观力的初步研究[J];机器人;2007年04期
4 赵娟平;;神经网络PID控制策略及其Matlab仿真研究[J];微计算机信息;2007年07期
5 王乐锋;荣伟彬;孙立宁;;纳米操作系统的研究现状与关键技术[J];制造技术与机床;2007年03期
6 田孝军;王越超;刘连庆;焦念东;董再励;席宁;;具有三维力反馈的原子力显微镜纳米操作系统[J];仪器仪表学报;2006年07期
7 李书义;;数码相机与传统相机的景深关系[J];科技情报开发与经济;2006年10期
8 嵇国金,马奎,王磊;微纳操作器运动学分析[J];微纳电子技术;2003年03期
9 罗安,路甬祥;专家PID控制器及应用[J];信息与控制;1992年03期
相关硕士学位论文 前3条
1 杨庆国;基于ARM和WinCE的手指静脉识别系统设计[D];哈尔滨工程大学;2011年
2 李彦林;基于OMAPL137嵌入式平台的手指静脉身份识别系统的研制[D];重庆理工大学;2011年
3 马贺;微/纳米二氧化锰的制备及其电磁特性的研究[D];大连理工大学;2009年
，本文编号：2339319
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2339319.html
