【作者】 吴笛飞；

【导师】 丁永生；

【作者基本信息】 东华大学 ， 模式识别与智能系统， 2008， 硕士

【摘要】 多自由度冗余驱动并联机器人控制中,其末端操作器位姿是反映并联机器人:运动状态的重要参数,并联机器人具有多自由度、运动轨迹复杂的特点,其位姿检测要求同时能完成多个自由度的测量,位姿检测是当前大载荷系统检测领域亟待解决的问题之一。相对于传统检测手段,计算机视觉检测系统具有非接触、智能、检测速度快等优点,具有较大灵活性,能实现多自由度测量,克服了传统检测手段在并联机器人位姿测量中遇到的困难。本论文从计算机视觉着手,研究构建视觉检测平台并通过不同时刻的图像智能化的采集场景中被测物体的运动信息,进而实现对多自由度冗余驱动并联机器人视觉检测系统的信息集成、处理。本论文建立了一种基于尺度不变量特征变换（SIFT）的视觉测量系统框架来检测一种具有10个输入的6自由度冗余驱动并联机器人,通过该系统检测出并联机器人的实时位姿。主要包括图像采集与传输、摄像机标定、SIFT匹配、空间点重建和位姿测量五个部分。首先根据冗余输入并联机器人位姿检测系统的特点,在硬件连接上使用两个工业黑白CCD相机WAT-902H经过同步锁相的方式连接一个PCI-1409多通道黑白图像采集卡,再将图像采集卡连接到PC机上,在尽量不影响系统性能的前提下降低系统复杂度和系统成本。在仿真平台搭建上,使用LabView和Matlab完成图像信息采集与显示系统的主要功能模块,以动态观测检测对象的状态。其次,为了简化计算提高系统的速度,采用较为简单的基于针孔模型的摄像机线性标定法标定对双目视觉系统进行标定,根据已知的样本点通过最小二乘法估算出基本参数。接着,在并联机器人位姿检测中的特征提取和特征匹配步骤中引入SIFT立体匹配算法,该算法尚未在该并联机器人位姿视觉检测系统中应用。匹配算法首先在尺度空间极值检测,剔除不稳定关键点之后确定关键点方向向量,生成SIFT特征向量,最后利用SIFT特征向量完成匹配。然后,通过点的重建方法将SIFT匹配算法得到的特征点2D坐标转换为特征点的实际3D坐标。最后,建立了冗余驱动并联机器人运动平台位姿计算的求解方法,将求解位姿问题转化为一个优化问题。通过前后两个时刻求得的特征点的实际3D坐标,根据三维空间的平移和旋转矩阵得出一组非线性方程,并将非线性方程组转化为函数优化问题,使用改进的PSO算法对该优化目标函数进行优化,针对优化函数的特点,改进的PSO算法主要加入了惯性系数,并且采用领域极值B_best代替G_best全局极值,减小了算法陷入局部极值的可能,在保证速度的前提下提高的精度,进而求出空间的实际位姿参数^{θ,（?）,φt_x,t_y,t_z}。仿真实验表明该算法在图像存在一定视角、平移、旋转、亮度和尺度的变化时仍然具有较好效果,证明了上述方法的有效性和可行性,适用于对并联机器人多自由度和空间复杂运动的检测。更多还原

【Abstract】 In the control of multi-degree of freedom (MDOF) parallel robot, the position and orientation of MDOF parallel robot is a very important motion state parameter. The MDOF parallel robot has complicated motion trails. Compared with tradition measurement methods, computer vision based measurement system has the advantage of intelligence, non-touch and more fast. The position and orientation measurement of MDOF parallel robot with high precision is a unresolved problem. In this thesis, the position and orientation measurement system for 6-DOF parallel robot is built based on machine vision theory. The sampling is actualized by images at different moments and then to intelligently extract the motion information from it. The position and orientation information of 6-DOF parallel robot is further analyzed and computed.This thesis presents a framework of scale-invariant feature transform based stereo vision position and orientation measurement system for a 6-DOF redundant actuation parallel mechanism with 10 inputs. The result of real-time pose can be calculated and displayed by the system. It is composed of image gather and transfer, camera calibration, scale-invariant feature transform (SIFT), reconstruct of points and position and orientation measurement. Main steps are as follows.First, a simple linear calibration method based on pin-hole model is used to calibrate the stereo vision position and orientation measurement system for parallel robot.Second, an acquisition and display simulation system of vision information based on LabView is built. It can observe the object state dynamically.Next, a processing system using SIFT is built. This framework is mainly based on scale-invariant feature transform. It performs reliable matching between different views of objects, when the images have change in scale, rotation, shift and change in illumination. It is especially adapt to measurement for multiple DOF parallel robot and complex motions in space. Finally, the parameter of pose can be calculated by matrixes of translation and rotation of 3D, and then is converted into an optimization problem. The result is optimized by improved particle swarm optimization (PSO) to improve the precision ensuring the high speed of the process. In the end, position and orientation measurement for parallel robot is realized by Matlab program.The results of simulation experiments prove the veracity and validity of all algorithms in this thesis.更多还原