【作者】 吴勇；

【导师】 郝矿荣；

【作者基本信息】 东华大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 大载荷对象的高精度智能控制不仅要求多阶段协同调控策略,更要求能对多运动目标进行实时距离检测、速度检测,来最终完成对多个运动目标的对接和控制。本次毕业设计就是基于这样的背景,设计一个基于二自由度机器人的视觉伺服系统。该系统是一个已标定的双目视觉系统,它采用遗传算法(GA),利用遗传算法自身的特点,能更快更好地找到全局最优点,从而得到较高匹配精度的点,最后通过三维重建和速度模型,实现对目标事物的运动距离、瞬时速度的动态检测,从而能实时监视并控制目标事物的运作。此检测系统的主要硬件设备是:两台东芝的外部参数完全相同的黑白CCD摄像机,加拿大Matrox公司的图像采集卡,PC机以及一个两自由度机器人。在软件方面,采用在Windows 2000或者Windows XP环境下都能运行的软件,它是MIL/ActiveMIL或MIL-Lite/ActiveMIL-Lite。第一步:通过CCD摄像头和采集卡,采集图像特征点。第二步:目标定位和跟踪:本系统针对两自由度机器人的背景颜色,选择与背景颜色有明显差异的黑色圆形螺丝作为检测标志物,同时选择机器人运动平台上的四个螺丝所组成的平行四边形为特征,实现特征提取,从而完成对特征点和运动目标的跟踪。第三步:经过对序列图像进行分析研究,通过运动学模型和几何模型,跟踪特征目标并同时计算出这个运动特征目标在连续帧图像中移动的欧式距离第四步:通过公式v=ds/dt,v=(?);s=(?),求出目标点的的旋转角速度。最后对论文进行了全面总结,并论述了一些对立体视觉检测技术的展望和此次毕业设计的体会。更多还原

【Abstract】 High-precision intelligent control of the Large load target requires not just cooperation control strategy of multi-stage, but also be able to get multi-targets’s real-time distance,speed,so as to complete multi-targets’s connecting and control finally.The graduation project is based of this background ,to design a visual servo system based of 2-DOF robot. The system is a calibration of the binocular vision system using genetic algorithms (GA), it uses of genetic algorithms own characteristics, be able to find the best overall points more faster and better, so can get higher accuracy of the match points and finally through the three-dimensional and speed of reconstruction to achive dynamic testing of the objectives’ moving distance, instantaneous speed, in order to be able to real-time monitor and control of the operation of targets . This detection system’s main hardware device: Toshiba’s black-and-white CCD camera, whose external parameters is same , Matrox’s image acquisition card of Canada, PC machines as well as a robot with two degrees of freedom.In terms of software, it uses software which can run in Windows 2000 or Windows XP environment ,it’s MIL/ActiveMIL or MIL-Lite/ActiveMIL-Lite.The first step: get images’ feature points through CCD camera and capture card.The second step: targeting and tracking: this system aims at the 2-DOF robot’ background color, and choose the round black screw as the detection symbol,which is significant different from the background color, at the same time, it choose parallelogram composed by four screws on the movement platform as the feature to realize feature extraction,in order to complete the feature points and moving target’ s tracking.Step three: after a series of images’ analysis, through kinematics and geometry model of models, track the feaure target, at the same time, calculate the target’s the movement distance in consecutive frames.Step four: Get the target’ s rotation angular velocity through these formula. Finally, a comprehensive summary of papers and discuss some prospect of stereo vision detection technology and experience of this graduate design.更多还原