An efficient omnidirectional vision system for soccer robots: From calibration to object detection

Robotic soccer is nowadays a popular research domain in the area of multi-robot systems. In the context of RoboCup, the Middle Size League is one of the most challenging. This paper presents an efficient omnidirectional vision system for real-time object detection, developed for the robotic soccer team of the University of Aveiro, CAMBADA. The vision system is used to find the ball and white lines, which are used for self-localization, as well as to find the presence of obstacles. Algorithms for detecting these objects and also for calibrating most of the parameters of the vision system are presented in this paper. We also propose an efficient approach for detecting arbitrary FIFA balls, which is an important topic of research in the Middle Size League. The experimental results that we present show the effectiveness of our algorithms, both in terms of accuracy and processing time, as well as the results that the team has been achieving: 1st place in RoboCup 2008, 3rd place in 2009 and 1st place in the mandatory technical challenge in RoboCup 2009, where the robots have to play with an arbitrary standard FIFA ball.     

激光视觉传感器的快速标定方法

建立了激光视觉传感器的数学模型,提出了一种分离参数对传感器结构参数进行标定的方法。结合标定要求,设计了标定靶。该标定方法装置简单,不需要对靶面坐标进行精确测量,简化了标定过程,计算速度快,能够保证较高的精度。试验表明,该方法是一种高效实用的视觉传感器标定方法。     

Calibration of geometric parameters and error compensation of non-geometric parameters for cable-driven parallel robots

Accuracy is an important performance indicator that affects the research and industrial application of cable-driven parallel robots (CDPRs). The error sources of CDPRs include geometric parameters (GPs) and non-geometric parameters (NGPs). Typically, GPs can be calibrated by external measurement devices whose position is dependent on coordinate system parameters. To improve the calibration accuracy and robustness, an iterative calibration method is proposed to calibrate the coordinate system parameters and GPs, and the asymptotic convergence is proven. Moreover, considering the tight coupling and non-linearity of NGPs, we design an artificial neural network to compensate for the residual position errors caused by NGPs. Based on the hierarchical genetic algorithm, a synchronization optimization algorithm is developed to improve the approximate accuracy and generalization to residual position errors of unknown trajectories. With theoretical initial parameters, experiments for the calibration of GPs and error compensation of NGPs were performed on a 3-DOFs CDPR. Finally, the average position error of the end-effector is reduced to 1.3 mm and the maximum error is 1.9 mm.     

A fast and accurate design method for broad omnidirectional bandgaps of one dimensional photonic crystals

A fast and accurate novel design method is proposed for the design of broad omnidirectional bandgaps of one dimensional (1-D) photonic crystals (PCs). Presented method is verified with various numerical examples for 1-D photonic crystals which consist of a cascade of two quasi-periodic stacks and broad omnidirectional bandgaps are achieved. Furthermore, computation time requirement of the presented method is considerably less than that required for purely numerical approaches. The proposed algorithm is quite flexible and can easily be modified to address problems involving 1-D PCs consisting of three and more cascaded stacks and specific 2D PC structures.     

HashHeat: A hashing-based spatiotemporal filter for dynamic vision sensor

Neuromorphic event-based dynamic vision sensors (DVS) have much faster sampling rates and a higher dynamic range than frame-based imagers. However, they are sensitive to background activity (BA) events that are unwanted. We propose HashHeat, a hashing-based spatiotemporal BA filter for DVS. It is the first spatiotemporal filter that doesn't scale with the DVS output size and doesn't store the 32-bits timestamps. We not only give the visual denoising effect of the filter but also use two metrics for quantitatively analyzing the filter's global performance and local performance respectively, where we introduce a novel metric for evaluating global performance. The experimental results show that HashHeat achieves similar global performance as baseline filters, but increases the signal to noise ratio by about 1.5x to nearly 5x compared with other baseline filters concerning the local performance. The hardware implementation enables HashHeat to output a labeled event every 10ns and meets the real-time requirement. And it can reduce the storage cost by 128x to 256x compared with baseline filters.     

结构光传感器在柔性视觉测量系统中的应用研究

针对柔性视觉测量系统的应用需求,根据介绍了基于线结构光视觉传感器实现棱线测量的相关技术,创新性地研究了采用线结构光视觉传感器实现空间圆孔测量的原理、方法,并对空间圆孔中心坐标的测量数学模型进行了详细地推导.基于线结构光视觉传感器实现棱线和空间圆孔测量的相关技术已经在江淮汽车厂C926汽车生产线上得到应用,采用双机器人形式对84个测量点进行了在线测量,取得了很好的测量应用效果.     

折/反射全方位视觉空间移变点扩散函数建模与图像复原

在分析折/反射全方位视觉(COV)图像模糊尺度分布规律的基础上提出了一种基于径向边缘检测的空间移变点扩散函数(SV-PSF)建模方法,并且研究了基于一种可以方便地加入空域可变模糊模型等先验知识的凸集投影(POCS)图像复原方法。本文提出的建模方法在有限的径向边界区域建立COV的全局SV-PSF模型,避免了一般建模方法需要大量的边缘检测和图像分割计算;采用基于POCS算法空域图像复原方法,很好容纳了SV-PSF这一先验知识,在实验中与MATLAB的盲反卷积工具箱函数的处理结果进行了对比,结果表明本文方法的灵活性和有效性。     

A novel ball handling mechanism for the RoboCup middle size league

This paper presents the hardware design and control design of a novel ball handling mechanism in the RoboCup Middle Size League used by team Tech United Eindhoven. The ball handling mechanism consist of two levers with two actively driven wheels attached to it, to exert forces on the ball in order to control its position relative to the robot. The proposed design is fully compliant to the rules and regulations imposed by the RoboCup Middle Size League community. The control design consists of a cascaded velocity and position feedback loop in combination with a feedforward controller which compensates for the robots ego motion. The proposed design is validated on a robot used by the Tech United Eindhoven team.     

A novel inverse dynamic model for 3-DoF delta robots

Delta Robots belong to a class of parallel robots widely used in industrial production processes, mostly for pick-and-place operations. The most relevant characteristics are the high speed and the extremely favorable ratio between the maximum payload and the weight of the robot itself. A reliable dynamic model is needed to implement torque controllers that reduce unnecessary high accelerations and so mechanical vibrations. The state-of-art inverse dynamic models exploit simplifications in order to facilitate the derivation of the equations of motion and their implementation. In this work, a novel and more rigorous inverse dynamic model is presented which does not rely on simplifications of the kinematic structure. The model has been validated comparing its estimations with real torques data collected moving a Delta Robot D3-1200 by SIPRO Srl; the computational complexity of the algorithm has also been investigated.     

光栅式双目立体视觉传感器的立体匹配方法

光栅式双目立体视觉传感器的难点之一在于立体匹配问题,为此,提出了一种基于极线约束和空间点最小距离搜索的立体匹配方法.该方法将光栅式双目立体视觉传感器看作两个光栅结构光传感器,分别标定后可测定光条中心点关于某个结构光模型的三维坐标,若两点匹配,则其三维坐标间的距离理论上为零.引入极线约束,在左摄像机成像光条上找一个特征点,在右摄像机所成像中便可计算出一条极线与之对应,在极线与各光条中心的交点中寻找匹配点.该方法在三维空间进行匹配,计算量小,能够实现点与点的唯一匹配.仿真实验表明了该方法的有效性.     

基于单目视觉的自主足球机器人自定位研究

本文介绍了一种在颜色场地信息已经由视觉识别的前提下,通过单目视觉测距,实现全自主足球机器人自定位的几何方法;对摄像机CCD像面的量化、镜头的非线性畸变和摄像机定标等可能的引起误差的因素进行了详细分析,并且对应地提出了解决办法.尤其是引入了偏最小二乘回归法对摄像机内部参数进行定标,避免由变量的相关性导致的定标误差;同时也更适于实际的比赛情况.实验结果表明它是可行的.     

A vision sensor with on-pixel ADC and in-built light adaptation mechanism

In this paper we propose an on-pixel Analogue-to-Digital Converter (ADC) based on pulse frequency modulation (PFM) scheme. This PFM based converter presents a viable solution for pixel-level based ADC. It uses a simple and robust circuit that can be implemented in a compact area resulting in a 23% fill-factor for a digital vision sensor in 0.25 μm CMOS technology. An in-built light adaptation mechanism has also been implemented which allows the sensor to better adapt to low-light intensity or to adjust the sensor saturation level. As a consequence, the proposed sensor features a programmable dynamic range. Image lag is eliminated since a reset of the photodetector is performed after the conversion period. A prototype comprising a 32×32 pixel array has been implemented in CMOS 0.25 μm technology. Each pixel occupies an area of 45×45 μm² with an average power consumption of 85 μW per pixel.     

A novel airflow sensor for miniature mobile robots

A novel airflow sensor has been developed for applications involving miniature chemical sensing robots. Information about air movement is essential for robots when they are searching for the sources of chemical plumes. The airflow sensor described here measures both wind direction and velocity at airflow rates commonly encountered in an indoor environment. Measurements are made by rotating a small paddle in the airflow. The varying speed of the paddle is analysed to determine both wind speed and direction. Low power consumption and rugged construction make the sensor well suited to robotic applications. The prototype sensor is small enough to fit on a mobile robot measuring only 10 cm in diameter. This paper presents the sensor operating principle, construction and some experimental results.     

一种面积紧凑的视觉芯片处理单元存储器

This paper presents a novel compact memory in the processing element （PE） for single-instruction multiple-data （SIMD） vision chips. The PE memory is constructed with 8×8 register cells, where one latch in the slave stage is shared by eight latches in the master stage. The memory supports simultaneous read and write on the same address in one clock cycle. Its compact area of 14.33 μm^2/bit promises a higher integration level of the processor. A prototype chip with a 64×64 PE array is fabricated in a UMC 0.18 μm CMOS technology. Five types of the PE memory cell structure are designed and compared. The testing results demonstrate that the proposed PE memory architecture well satisfies the requirement of the vision chip in high-speed real-time vision applications, such as 1000 fps edge extraction.     

立体视觉测量中摄像机标定的新技术

本文实现并改进了一种利用径向排列约束的摄像机标定算法,并利用该算法标定了安装在ＺＥＩＳＳＣＭＭ坐标测量机伺服机构上的ＣＣＤ摄像机的内部和外部参数,算法巧妙地利用了成像过程中的ＲＡＣ分解摄像机参数,提出了带有一阶径向畸变的摄像机模型,利用合理组织的求解次序使得每一步参数的求解都只须求解线性方程组,从而彻底改变了以往摄像机标定依赖于非线性方法。     

一种基于机器视觉的铝管缺陷检测系统

针对铝管生产过程中对准确、可量化的自动缺陷检测系统的迫切需要,本文引入一种由图像采集、缺陷检测、缺陷处理等模块组成的铝管缺陷检测系统。平板探测器获取由X光高压电源产生,穿过铝管的X射线并把所形成的数字图像通过USB端口发送至检测服务器。检测服务器使用机器视觉算法检测图像中的缺陷。当服务器检测到缺陷时,会向PCI板上指定位输出信号,报警装置接到信号后报警提醒工作人员。实验表明该系统能够自动、准确的标记出铝管中存在的缺陷,达到了系统的设计目标。     

A machine vision system for inspection of keyboards

A machine vision system for automatic inspection of keyboards is described. The system verifies that each location on a keyboard contains the correct, properly oriented key and that the graphics are not badly distorted. The graphics are inspected on the top and front surface of keys that vary widely in color, contour, surface texture and graphic content. Economical solutions are presented to a number of problems related to the design of the system's mechanics, optics, illumination, data reduction, pattern location and graphics discrimination. Operating procedures for training and running the system are also presented along with preliminary results.     

一种基于图像处理的双目视觉校准方法

双目视觉是利用机器视觉进行障碍物检测的研究热点。针对双目视频不同步,导致立体匹配不精准的问题,提出了一种基于图像处理的双目校准算法。算法首先根据道路的先验特征模型,建立视觉校准的敏感区域,以减小计算量。然后对图像的灰度进行聚类,并通过边缘检测和形态学处理来检测道路中的目标物,从而确定动目标与固定物的距离,搜索同步视频帧,实现双目校准。室外真实场景的实验结果表明,该算法可以较好地校准双目视觉,从而使立体匹配更精准。     

A robust and energy-efficient data dissemination framework for wireless sensor networks

Wireless sensor networks (WSNs) are appealing in obtaining fine-granular observations about the physical world. Due to the fact that WSNs are composed of a large number of low-cost and energy-constrained sensor nodes, along with the notorious time-varying and error-prone nature of wireless links, scalable, robust and energy-efficient data dissemination techniques are needed for the emerging WSN applications such as environment monitoring and surveillance. In this paper, we examine this emerging field from the point of view of supply chain management and propose a hybrid data dissemination framework for WSNs. In particular, for each sensing task, the whole sensor field is conceptually partitioned into several functional regions based on the supply chain management methodology. Different routing schemes are applied to different regions in order to provide better performance in terms of reliability and energy consumption. For this purpose, we also propose a novel zone flooding scheme, essentially a combination of conventional geometric routing and flooding techniques. Our hybrid data dissemination framework features low overhead, high reliability, good scalability and flexibility, and preferable energy efficiency. Detailed simulation studies are carried out to validate the effectiveness and efficiency of our scheme.     

基于双目视觉的光束法平差新算法

针对双目视觉中两摄像机的固定约束关系,提出了基于双目视觉的光束法平差新算法.对传统光束法平差法方程构造过程进行了改进,使每次迭代过程中只需对拍摄的两幅图像中一个摄像机的外参数进行优化,更好地将两个摄像机绑定为一个摄像机,使优化精度更高;新算法中,法化矩阵的维数比传统算法降低了1/2,所以每次迭代计算所需要的时间也会减少...