基于粒子滤波的移动机器人SLAM改进算法

针对未知环境中移动机器人同时定位和地图创建（Simultaneous Localization and Map Building,SLAM）由于机器人位姿和环境地图都不确定导致定位和地图创建变得更加复杂,提出一种局部最优（全局次优）参数法,即通过局部最优的位姿创建局部最优的环境地图,再通过局部最优的环境地图寻求局部最优的位姿,如此交替进行,直到得到全局确定性的位姿和确定性的环境地图。实验结果表明,同标准的基于粒子滤波的SLAM 算法（Particle Filtering-SLAM,PF-SLAM）比较,改进的算法提高了机器人SLAM过程中定位的准确度和地图创建的精确度,为机器人在未知的室外大环境同时定位和地图创建提供新的方法。     

一种改进的粒子滤波SLAM算法*

提出一种改进的粒子滤波SLAM（simultaneous localization and map building）同时定位和地图创建实现方法。改进方法让机器人大约行进10步完成基于局部已创建地图下的粒子滤波定位后,再利用激光传感器探测环境并更新创建的地图;同时在利用粒子滤波定位时,使粒子只分布在由航位推算法得出的机器人位姿附近,从而可有效地减少粒子的数量。实验结果表明,与标准的粒子滤波SLAM 算法比较,改进算法提高了机器人SLAM过程中定位和地图创建的精度和实时性,并为移动机器人在室外未知环境同时定位和地图创建提供了新方法。     

基于激光雷达的移动机器人定位和地图创建

针对未知环境下移动机器人定位和地图创建问题,提出了创建局部地图-机器人定位-更新全局地图的方法和步骤.利用激光雷达数据,采用"聚合-分割-聚合"的方法并运用动态阈值获得精确的局部地图.通过匹配局部地图和全局地图的线段关系,实现了机器人的定位,更新了机器人的位姿,减少了机器人的系统误差和环境误差,进而完成了全局地图的创建与更新.实验证明,此算法可以在实现机器人定位的同时获得精确的环境地图,并且有效的缩短了地图生成时间,可靠性高、实时性好.     

自主移动机器人的室内结构化环境地图创建

定位与地图创建是自主移动机器人领域研究的重要课题.本文阐述了一种以扩展卡尔曼滤波算法为主要框架,运用直接位姿控制模型描述机器人运动的算法,实现了机器人在室内结构化环境中的同时定位和地图创建.仿真与实验结果表明,里程计信息无法满足定位和创建环境地图的要求,本文算法则能够实现机器人的精确定位.并生成满足一致性要求的地图.     

煤矿巡检机器人同步定位与地图构建方法研究

针对煤矿井下无GPS环境下巡检机器人自主定位问题,研究了基于激光雷达的同步定位与地图构建方法。首先建立激光雷达观测模型和里程计预测模型,将机器人定位和地图构建的实际问题转换为概率数学模型的逻辑推理问题。同时采用自适应蒙特卡罗定位算法进行机器人实时位姿估计,提出了根据粒子权重(地图的匹配度)进行重采样的方法,以去除权重小的粒子,实现了用较少、较好粒子精确表达机器人位姿的后验概率分布,满足机器人利用传感器在栅格地图上实时定位的需求。通过对Fast-SLAM算法进行优化,减少了粒子数量,缓解了粒子耗散,提高了地图构建的精确性。实验结果表明,基于激光雷达的同步定位与地图构建方法有效解决了巡检机器人实时位姿估计和环境地图构建的问题,结合自适应蒙特卡罗定位算法和优化Fast-SLAM算法提高了机器人定位的自适应性和地图构建的精确性。     

基于ROS平台的移动机器人自主导航技术研究

针对移动机器人传统的室内环境导航方法比较单一,无法适应多场景等问题,本文将传统的基于ROS平台的Navigation功能框架的导航方法与AprilTag标签算法相结合进行导航系统的设计。该系统采用Gmapping算法和AMCL算法分别完成环境建图与机器人定位,然后通过Navigation功能框架获取环境地图与机器人位姿信息进行路径规划,并且在此基础上结合AprilTag标签算法进行机器人的导航设计。实验结果表明,该系统构建的环境地图精度较高,能够实现对机器人的有效定位,且Navigation功能框架与AprilTag标签相结合的导航系统能更好地适应多场景。     

基于多传感器信息融合的机器人导航与定位技术研究

移动机器人已经服务于各个领域,作为室内移动机器人服务的前提是实现机器人精准定位。文章介绍了一种射频识别（Radio Frequency IDentification,RFID）和即时定位与地图构建（Simultaneous Localization and Mapping,SLAM）多传感器融合库存盘点机器人精准导航与定位的方法。通过读取布置在环境中的RFID标签,获取机器人的大概位置信息和准确运动状态,然后结合SLAM技术实现精准定位。实验结果表明,该方法可以使库存盘点机器人精准定位,且系统搭建简单、成本低,对室内机器人定位具有参考价值。     

基于Kinect和视觉词典的三维SLAM

针对传统滤波器方法解决机器人同时定位与地图创建(SLAM)时的误差积累问题,提出了一种基于视觉词典(BOW)的三维SLAM算法,以有效解决机器人长时间运动下误差积累的问题。相比图优化SLAM中常用的随机检测和Kd树(Kd-Tree)算法,采用基于树结构的视觉词典闭环检测算法来提高相似场景的检索效率。首先采用基于GPU的特征提取算法提取图像特征,并利用交叉匹配和k最近邻(kNN)算法取得图像中鲁棒性较强的内点;然后通过基于随机抽样一致性奇异值分解(RANSAC SVD)算法计算出相邻帧的初始位姿变换,并利用通用迭代最近点(G-ICP)算法进行优化,得到高精度的位姿变换;最后利用增量平滑和建图(iSAM)图优化方法得出最终位姿,拼接出高精度的点云地图和运动轨迹。标准数据集的测试表明,所提算法在复杂情况下具有良好的鲁棒性和精度。     

融合语义激光与地标信息的SLAM技术研究

针对移动机器人在SLAM（即时定位与地图构建）过程中出现的定位失真问题,提出一种通过搭建地标数据库和位姿推导模型,修正机器人错误定位的方法。建图过程中,融合视觉信息与激光数据,得到语义激光,赋予地标语义标签并记录其在地图上的位置信息。导航过程中,当产生定位偏差时,结合多种位姿数据和相对位置关系,推算出机器人在地图上的实际位置,完成重定位。通过实验测试可知,该方法克服了现有机器人在实际室内动态环境下,单一地采用激光或视觉进行定位或重定位技术的缺点和不足,能有效解决“机器人位置漂移问题”。将机器人从当前位置劫持到另一位置,也能根据提出的算法迅速重定位,且定位精度高。     

基于扫描匹配的室外环境SLAM 方法

针对室外环境中的机器人“绑架”问题,提出了基于地图匹配的SLAM方法．该方法舍弃了机器人里程计信息, 只利用局部地图和全局地图的图形相关性进行机器人定位．方法的核心是多重估计数据关联,并将奇异值分解应用到机器人位姿计算中．利用Victoria Park数据集将本算法与基于扩展卡尔曼滤波器的方法进行比较,实验结果证明了本文提出的算法的有效性．     

Indoor robot localization based on single RFID tag

With the mature of the RFID technology and the wide use of the Internet of Things technology, it is a new research hotspot that RFID technology was applied to robot localization and navigation. RFID tags have the characteristic of storing information, long reading distance and using conveniently. A novel robot localization method combining RFID with laser data is proposed in this paper. When the robot detects an RFID tag, RFID antenna angle probability model is determined based on the fuzzy reasoning algorithm. Then the Bayes rule is used to estimate the direction of the robot. Finally, the robot global localization is realized by fusion of the laser ranging data. The experimental results showed that the precision of the positioning based on single RFID tag is enhanced greatly by combining laser detecting data.     

基于RFID技术的机械手零件管理

结合现行RFID技术的应用现状提出了基于RFID技术的机械手零件生产管理.首先阐述了RFID技术的原理、RFID系统的基本工作流程、RFID系统的选择.在此基础上,采用PRO/ENGNEER软件对平行限位式机械手部进行了设计,手部零件主要由手指、指套、导块等组成.然后,从三个层面:机械手零件生产流水线上的管理、机械手零件的仓库管理、机械手零件在供应链中的管理等方面做了初步的探讨,说明了基于RFID技术管理机器人零件的优越性.     

Assistive robot application based on an RFID control architecture and a wireless EOG interface

This paper describes an assistive robot application that combines a portable wireless interface based on electrooculography (EOG) and Radiofrequency Identification (RFID) technology. This assistive application is aimed at handicapped users who suffer from a severe motor disability. To that end, a realistic application has been designed. It consists of an environment in which users can bring a glass and a water bottle closer with only the help of their eye movement using a real robot arm. RFID will be used as a support to the EOG interface in a shared control architecture by storing information of the objects in tags placed on the scene. Five volunteers tested the assistive robot application. The results obtained show that all of them were able to finish the tests in a suitable time and the results improved with practice and training. This proves that the assistive robot application can be a feasible way to help handicapped users.     

基于RFID技术的室内机器人定位方法的研究

近年来,随着RFID技术的日益成熟,它在机器人领域的应用变得越来越广泛而深入。文章首先分析了RFID技术在室内环境下应用于移动机器人自我定位的可行性,然后描述了使用蒙特卡罗概率统计模型对机器人附近的RFID标签进行方位定位的过程,并且阐述了利用在此过程中获得的数据对自身或者特定物在特定环境下进行方位定位的方法,最后文章分析了此技术在某些特定场合进行应用推广的可行性和前景。     

Shared control architecture based on RFID to control a robot arm using a spontaneous brain–machine interface

This paper describes a shared control architecture combining a Brain–Machine Interface (BMI) with Radio-frequency Identification (RFID) technology to control a robot arm in pick and place operations. A non-invasive spontaneous BMI capable of distinguishing between three different mental tasks has been designed. Using the BMI, the user can control the robot in order to perform complex actions (e.g. pick and place operations). RFID tags have been placed in the experimental setup to give information about the position of the objects in the scene. With this information, the user is able to pick and place the objects with a robot arm by performing simple commands: move left, move right, pick or place, with the only help of the BMI. Four volunteers have successfully controlled the robot arm, and time and accuracy have been measured.     

RFID和WSNs技术在机器人协同定位中的应用研究

针对常用定位技术难以满足移动机器人协同定位精度高、实时性强的要求,提出了射频识别(RFID)和无线传感器网络(WSNs)混合定位技术。采用有源RFID定位技术,参考身份标识ID对应的电子标签在未知环境中的位置坐标,利用接收到的信号强度结合改进的极大似然估计算法确定出单个机器人的位置坐标。基于WSNs技术,确定出父节点机器人与预协作子机器人之间的相对方向和距离信息。给出了机器人自身绝对位置和协作机器人之间相对定位的算法,并通过实际电路和程序设计进行了实验验证。结果表明:该方法有效,可取,可解决多机器人可协同定位的范围小,精度低,实时性差等问题。     

An approach for research and training in enterprise information system with RFID technology

Successful implementation of Radio Frequency Identification (RFID) in Enterprise Information System (EIS) depends on exploring the knowledge of RFID technology and providing EIS software training to potential users. The restriction of RFID technology and huge amount item-level data collection via RFID potentially increases the complexity of integrating with EIS. It raises several concerns such as reduced data accuracy, and increased exceptions leading to processes requiring more time to be developed, tested and maintained. However, this approach has not been entirely investigated so far. This paper proposes an integrated research on EIS and a training approach for students and employees who intend to integrate RFID with EIS software in an organization. This integrated research approach applies a typical EIS project workflow from automatic shop floor data collection to customer order fulfillment. This research identifies these stages as transaction, planning and scheduling, integration, execution and control, and automatic identification and data collection. For training and demonstration purpose, this research proposes prototype information systems that emulate the systems used in any organization. This ensures that the logic of the proposed methodology of integrating RFID and EIS software is consistent with that in the business organization. This paper discusses some of the important case studies that were conducted with the suggested prototype systems.     

基于RFID和组态王的停车场管理系统

介绍了一种可以实现无人值守,远程监测的停车场智能管理系统.将RFID技术、自动控制技术、计算机技术及组态软件应用到停车场的管理中,实现对车辆出入控制,停车位远程监测的现代化管理.监控设备、出入口门禁系统和停车位车辆检测器的信号通过RS-485总线与中心管理计算机通信.利用组态软件设计远程监测画面对停车位进行远程监测、调度.     

Direction Sensing RFID Reader for Mobile Robot Navigation

A self-contained direction sensing radio frequency identification (RFID) reader is developed employing a dual-directional antenna for automated target acquisition and docking of a mobile robot in indoor environments. The dual-directional antenna estimates the direction of arrival (DOA) of signals from a transponder by using the ratio of the received signal strengths between two adjacent antennas. This enables the robot to continuously monitor the changes in transponder directions and ensures reliable docking guidance to the target transponder. One of the technical challenges associated with this RFID direction finding is to sustain the accuracy of the estimated DOA that varies according to environmental conditions. It is often the case that the robot loses its way to the target in a cluttered environment. To cope with this problem, the direction correction algorithm is proposed to triangulate the location of the transponder with the most recent three DOA estimates. Theoretical simulation results verify the reliability of the proposed algorithm that quantifies the potential error in the DOA estimation. Using the algorithm, we validate mobile robot docking to an RFID transponder in an office environment occupied by obstacles.     

无线射频识别在邮政速递总包处理中的应用

无线射频识别技术是利用大规模集成电路与无线通信技术相结合的自动识别和数据采集技术,代表着自动识别领域的发展方向。论文将这种技术应用到邮政速递的总包处理过程中,阐述了RFID在速递生产应用的基本框架、频率选择、信息标识、工艺流程、信息系统与数据组织以及射频袋牌回收调拨等完整的解决方案,并对实施效果进行了分析。