基于四元数最少特征点的相机位姿估计

为解决相机位姿估计的问题,提出了一种基于四元数最少特征点的相机位姿估计算法。在相机拍摄的二维图像中检测并匹配特征点,根据特征点坐标与约束条件建立多项式系统,通过求解对应的矩阵方程来求解多项式系统。用四元数表示相机的旋转,避免了求解中相机旋转与平移相互纠缠的问题,当2个摄像机视图之间的平移为零时能够很好地进行位姿求解。求解中结合5点算法对求解原理进行了详细推导,并进行抗噪声测试。测试中随着匹配的特征点数增加,算法平均估计误差范围不超过2%。利用KITTI数据集测试算法的实用性,并与其他几种方法进行结果比较。结果显示,相同条件下算法将估计精度提高了24.5%以上,体现出良好的估计精度。     

机器人手眼矩阵求解算法

机器人手眼矩阵计算是机器人视觉、机器人校准以及视觉伺服等研究领域共同的课题 ,目前主要通过不同的数学工具如四元数、对偶四元数、旋量来描述空间运动 ,或者施加不同的运动约束如纯移动或纯转动来将旋转和平移解耦开来。文中利用矩阵直积和矩阵特征向量推导了机器人手眼转换矩阵的线性方程 ,无需特殊的数学工具而且有效解决了Nicolas舍弃方程引入的误差 ,并通过最小二乘法得到线性闭解 ,简化了计算并提高了效率。采用Rodrigues公式对所求解的旋转部分进行正交化以消除测量噪声的影响 ,由计算试验证明正交化所引入的误差在绝大多数情况是允许的     

大部件对接中基于单位四元数的iGPS测量位姿比对研究

在大尺寸飞机部件数字化对接过程中,iGPS的应用越来越广泛.本文在研究iGPS测量数据结构与形式的基础上,提出了基于单位四元数的iGPS测量位姿比对算法.与目前常用的一些位姿比对方法相比,该算法具有计算简洁明了、编程实现便捷的优点,尤其适用于大旋转角、多点实时测量的情形.验证实验结果表明,在不同旋转角和平移量的情况下,该算法均具有很高的转换精度,可有效适用于iGPS的测量位姿比对.     

直升机光电助降系统的双摄影测量算法及实验研究

基于信标的摄影测量定位技术,提出了一种用于光电助降系统的双摄影测量算法.利用两台CCD 摄像机探测直升机两侧各四个信标,建立目标坐标系、摄像机坐标系和图像坐标系,根据信标、镜头中心和信标像点的共线性得到关于旋转矩阵和平移矢量的线性方程组,由此计算出直升机相对于着降点的位姿.设计了一套直升机光电助降实验系统,使用高性能DSP 系统进行实时图像处理,运用双摄影测量算法实时计算直升机着降引导信息.模拟实验结果表明,在两摄像机距离着降点8～13 m 范围内,光电助降实验系统的距离测量精度达到±5 cm,而角度测量精度达到±3°.     

利用共面圆特征实现增强现实图像注册

本文提出以对象上的共面圆为基础估算对象位姿,实施增强现实图像注册的方法.假设相机内参数已标定,文章从绝对二次曲线与平面圆有公共虚交点出发,通过计算消隐线方程、确定圆心位置、获取坐标轴方向等步骤分另估算对象位姿的旋转分量和平移分量,从而最终估算出对象位姿.该算法所需预知的条件少,计算代价小,因而在工业环境中更易实现.综合数据仿真与真实对象测量验证了本文算法的正确性以及在噪声和遮挡情形下的鲁棒性.     

一种新型非接触位姿检测系统研究

提出一种基于位置敏感探测器(简称PSD)多光束同步检测技术的位姿检测新方法,采用两个二维PSD从不同角度同时测量出目标物体上多个特征点的二维坐标,然后通过坐标变换计算得到各点的三维空间坐标;根据位姿变化前后的各点空间坐标,采用比较简单的算法计算出目标物体的位姿参数.通过算例验证该方法是可行的;由初步的实验获得目标物体的位姿参数,最大平移误差为1.0574mm,最大旋转误差为0.2851rad,需要通过多种途径进一步提高系统的检测精度.     

不规则零件优化排样的神经网络混合优化算法

 提出一种利用人工神经网络求解不规则件排样问题的混合优化方法.该方法首先把排样和制造工艺联系起来,将多边形各边向外扩充,为零件预留加工余量;然后采用自组织特征映射模型（SOM）和Hopfield人工神经网络相结合的方法,运用SOM神经网络对初始在板材内随机排布的不规则零件进行平移,逐步减小不规则零件之间的重叠面积,求得各零件的最优位置,再运用Hopfield神经网络对平移后的零件旋转,进行迭代运算,当能量函数达到稳定状态时,得到各排样零件的最优旋转角度组合,实现自动排样.算法可以解决不规则件和矩形件在规则板材以及不规则板材上的排样问题,实例证明了该算法的有效性和实用性.     

船用光电系统中摄像机参数的标定方法

基于一阶径向畸变的针孔摄像机模型，提出了一种新的线性三步摄像机定标方法，即通过径向排列约束计算摄像机参数的旋转矩阵、X轴平移矢量和Y轴平移矢量；根据透视投影的交比不变性解算一阶径向畸变参数；利用求得的摄像机参数建立有效焦距和Z轴平移矢量的线性方程，采用最小二乘法得到线性解。实验表明，该方法简单快捷，具有较高的标定精度，解决了原有算法采用非线性搜索寻优可能存在的解不稳定性等问题。     

求解四元数矩阵方程AHXA=B的基于矩阵半张量积的新方法

四元数线性系统在控制理论和工程中有广泛的应用。利用矩阵半张量积对四元数矩阵方程进行研究，提出四元数矩阵的一种实向量表示并研究其性质。结合实向量表示与矩阵半张量积，给出四元数矩阵方程A^HXA=B的极小范数Hermitian解的存在条件及通解表达式，并且给出相应算法。数值实验证明了实向量表示方法的可行性。     

用矩阵微分建立股骨颈手术导航机器人位姿误差模型的方法

应用矩阵微分建立了股骨颈手术导航机器人的位姿误差模型.依次做了以下主要工作:设计股骨颈手术导航机器人的三维形体结构,即串、并联混合机构;基于结构与运动特征,采用传递矩阵建立手术导航模型;提取特征参数作为微分变元,通过矩阵微分建立手术导航的位姿误差模型;应用仿真软件matlab7.0,以导航机器人的设计尺寸与容差值为变量,对特征参数引起的导航位姿误差分布进行仿真,其结果位姿误差呈平面分布,极大误差在边界线上获得.应用矩阵微分建立的手术导航误差分析模型的工程含义明确,结构规范,适用于将机器人的精度校核穿插到形体设计的前期阶段进行,在并行设计中的精度校验上有实用性.     

Conservative integration of rigid body motion by quaternion parameters with implicit constraints

An angular momentum and energy‐conserving time integration algorithm for rigid body rotation is formulated in terms of the quaternion parameters and the corresponding four‐component conjugate momentum vector via Hamilton's equations. The introduction of an extended mass matrix leads to a symmetric set of eight state‐space equations of motion. The extra inertial parameter serves as a multiplier on the kinematic constraint, and it is demonstrated that convergence characteristics are improved by selecting this parameter somewhat larger than the inertial moments. External loads enter these equations via the set of momentum equations. Initially, the normalization of the quaternion array is introduced via a Lagrange multiplier. However, this Lagrange multiplier can be expressed explicitly in terms of the gradient of the external load potential, and elimination of the Lagrange multiplier from the final format leaves only an explicit projection applied to the external load potential gradient. An algorithm is developed by forming a finite increment of the Hamiltonian. This procedure identifies the proper selection of increments and mean values, and leads to an algorithm with conservation of momentum and energy. Implementation, conservation properties, and accuracy of the algorithm are illustrated by examples with a flying box and a spinning top. Copyright © 2012 John Wiley & Sons, Ltd.     

Dynamic calibration of the relative pose and error analysis in a structured light system

We investigate the problem of dynamic calibration for our structured light system. First, a method is presented to estimate the rotation matrix and translation vector between the camera and the projector using plane-based homography. Then an approach is introduced to analyze theoretically the error sensitivity in the estimated pose parameters with respect to noise in the projection points. This algorithm is simple and easy to implement. Finally, some numerical simulations and real data experiments are carried out to validate our method.     

基于深度学习算法的矿用巡检机器人设备识别

为了实现矿用巡检机器人对煤矿井下设备的识别与匹配,通过基于卷积神经网络的深度学习算法建立了煤矿设备类型识别模型,分别在明亮环境下、昏暗环境下以及设备重叠情况下采集大量待识别设备图像样本,再对识别模型进行训练,实现巡检机器人对煤矿设备的精确识别与分类。使用基于粒子群优化的SVM（support vector machine,支持向量机）建立了煤矿设备匹配模型,将巡检机器人相对于煤矿坐标系的三轴位置信息、三自由度角度和视觉相机转角作为匹配模型的输入量,将相机视野中设备序号作为输出量,实现煤矿设备类型识别模型识别出的设备与已知设备序号一一对应。实验结果表明基于深度学习算法的煤矿设备类型识别模型对外界的干扰不敏感,识别准确率高;基于SVM的煤矿设备匹配模型的匹配准确率达到了93.2%,在匹配准确率的训练和测试效率上均优于基于BP（back propagation,反向传播）神经网络的匹配模型。研究结果可为煤矿井下巡检机器人的研制提供参考。     

三维点云的两步校准法及其应用研究

针对具有平面特征的工件的位姿识别,提出了一种三维点云两步校准法。首先,对三维点云进行多次随机抽样,每次抽取3个点并计算其所构成平面的单位法向量;经多次抽样后,得到一个平面单位法向量集合,采用密度聚类算法提取主平面的单位法向量,并根据主平面单位法向量与目标向量的轴角变换关系实现点云的倾斜度校正。然后,先将点云投影至主平面,再将各点的高度映射为主平面内的灰度,得到点云灰度图;再利用图像模板匹配算法获取点云在主平面内投影的平移量和旋转角,经平移和旋转后实现点云位姿的最终校准。将所提出的三维点云两步校准法应用于高压输电线塔塔座自动焊接系统,以识别塔座的位姿信息,并与基于采样一致性法识别的塔座位姿信息进行对比。结果表明,相比于采样一致性法,所提出的两步校准法既提高了工件位姿的识别精度,又缩短了约76%的耗时。研究结果可为具有平面特征的工件的位姿快速校准及识别提供参考。三维点云两步校准法在先进制造业中具有广阔的应用前景。     

基于视觉标定的包装搬运机器人定位方法

目的为了提高包装搬运机器人的定位精度,提出一种基于机器视觉的末端执行器定位方法。方法基于OpenCV设计一种视觉标定算法,该算法包括摄像机标定和位姿标定,可实现待码放物体图像坐标和机械手坐标之间的变换。结合工控机和运动控制卡设计其控制系统,同时给出硬件设计和软件设计方法。最后进行实验研究,包括原点定位和重复定位。结果实验结果表明,所述控制方法能够提高搬运机器人的定位精度,原点定位误差约为0.14 mm,重复定位误差约为0.6 mm。结论该搬运机器人定位方法能够满足包装码垛要求。     

Research on Multi-View Image Reconstruction Technology Based on Auto-Encoding Learning

Traditional three-dimensional (3D) image reconstruction method, which highly dependent on the environment and has poor reconstruction effect, is easy to lead to mismatch and poor real-time performance. The accuracy of feature extraction from multiple images affects the reliability and real-time performance of 3D reconstruction technology. To solve the problem, a multi-view image 3D reconstruction algorithm based on self-encoding convolutional neural network is proposed in this paper. The algorithm first extracts the feature information of multiple two-dimensional (2D) images based on scale and rotation invariance parameters of Scale-invariant feature transform (SIFT) operator. Secondly, self-encoding learning neural network is introduced into the feature refinement process to take full advantage of its feature extraction ability. Then, Fish-Net is used to replace the U-Net structure inside the self-encoding network to improve gradient propagation between U-Net structures, and Generative Adversarial Networks (GAN) loss function is used to replace mean square error (MSE) to better express image features, discarding useless features to obtain effective image features. Finally, an incremental structure from motion (SFM) algorithm is performed to calculate rotation matrix and translation vector of the camera, and the feature points are triangulated to obtain a sparse spatial point cloud, and meshlab software is used to display the results. Simulation experiments show that compared with the traditional method, the image feature extraction method proposed in this paper can significantly improve the rendering effect of 3D point cloud, with an accuracy rate of 92.5% and a reconstruction complete rate of 83.6%.     

Efficient Lagrangian relaxation algorithms for industry size job-shop scheduling problems

We improve the job specific decomposition Lagrangian relaxation algorithm applied to industry size job shop scheduling problems with more than 10 000 resource constraints. We introduce two new features in the Lagrange multiplier updating procedure. First, the usual solution of all subproblems followed by dual cost estimation and update of multiplier values is replaced by the estimation of a surrogate dual cost function and a more frequent update of multipliers is implemented after each subproblem solution. Second, an adaptive step size in the subgradient based multiplier update is introduced. Asymptotic properties of the surrogate dual cost function are obtained and the proposed algorithmic improvements are evaluated in extensive numerical examples including published data used by other researchers, as well as extensive real industrial scheduling system data.     

等效旋转矢量法在旋转弹姿态解算中的应用

作为捷联惯导的核心问题,姿态更新算法的选择至关重要.针对旋转弹的姿态测试问题,在四元数增量算法的基础上利用等效旋转矢量法对姿态进行优化,给出了姿态解算的数学模型.仿真结果显示,该算法对不可交换误差的抑制效果明显,同时,在一定程度上减小了累积误差,提高了导航精度.     

An augmented Lagrange programming optimization neural network for short-term hydroelectric generation scheduling

An approach based on augmented Lagrange programming neural networks is proposed for determining the optimal hourly amounts of generated power for the hydro-units in an electric power system. This methodology is based on the Lagrange multiplier theory in optimization and searches for solutions satisfying the necessary conditions of optimality in the state space. The equilibrium point of the network satisfies the Kuhn–Tucker condition for the problem. The equilibrium point of the network corresponds to the Lagrange solution of the problem. The proposed technique has been applied to a multi-reservoir cascaded hydro-electric system with a non-linear power generation function of water discharge rate and storage volume. The water transportation delay between connected reservoirs is also taken into account. Results obtained from this approach are compared with those obtained from the two phase optimization neural network and the conventional augmented Lagrange multiplier method. It is concluded from the results that the proposed method provides better results with respect to constraint satisfaction and is very effective in yielding optimal hydro-generation schedules.