摘要
在自主移动服务机器人系统中，获取当前机器人的准确位置(即定位)是完成导航和追踪等任务的前提。WLAN定位技术和RFID定位技术是新兴的定位技术，具有定位精度高、低功耗、低成本等特点，适合于移动机器人的定位。目前，在定位系统中通常采用单一传感器获取信息，这种方法只能获得环境特征的部分信息，导致系统定位精度低。因此，本文提出采用WLAN和RFID建立多传感器信息融合系统应用于移动机器人定位系统中以提高系统定位精度。论文以国家863计划项目“基于多信息融合的移动机器人自主定位关键技术研究”为支撑，对WLAN和RFID建立定位系统中的若干关键问题进行了研究，为多信息融合的研究奠定了基础。本文具体的研究工作如下所述：
在WLAN无线定位系统中，针对移动机器人系统定位特点，分析无线定位技术特点，研究无线网络定位算法，确定使用基于RSSI值的定位方法，分析了系统路径损耗模型。根据实验环境，确定信号传播模型，进行仿真实验并分析实验结果。利用无线路由器和无线网卡建立WLAN移动机器人无线定位系统，研究基于信号强度(RSSI)值的WLAN定位方法。通过多组测试实验，确定三边定位法作为WLAN定位系统的定位算法。
在RFID定位系统中，针对移动机器人室内定位系统环境复杂、障碍物多的特点，采用有源射频模块，基于Visual C++6.0平台开发了一种室内移动机器人定位系统。该系统中，定位标签选用RSSI值最大的四个标签，在极大似然估计法的基础上，引入移动误差因子，提出改进的极大似然估计定位算法，并将其应用在移动机器人定位系统中。
在联邦滤波融合结构中采用无迹卡尔曼滤波(UKF)对子滤波器进行滤波，针对UKF技术易发散、计算速度慢和采样的非局部效应问题，提出基于最小偏度采样和衰减记忆平方根滤波的UKF算法和采用比例修正系数最小偏度采样的UKF算法，通过引入最小偏度单形采样策略减少了采样点数，以衰减记忆法滤波代替协方差阵进行递推运算，减少了计算误差引起的滤波发散现象，将比例修正系数引入最小偏度采样策略中的算法，有效地解决了采样的非局部效应问题。为提高融合系统的定位精度和定位速度奠定了基础。
在WLAN和RFID定位系统的基础上，考虑使用单一的传感器不能满足定位精度要求，提出将联邦卡尔曼滤波技术应用在定位系统中，融合分别由WLAN、RFID定位系统获得的数据，实现整体精度的提高。在融合计算过程中因为局部滤波器有全局滤波器的反馈重置，这样，局部滤波器的精度也得以提高。进行了实验结果的比较，验证了所提出算法的可行性。
通过本文的研究，利用WLAN/RFID建立融合定位系统，可降低系统的成本、提高定位效率、改进设计质量、提高市场竞争力，具有重要的理论意义和学术价值。
In the system of autonomous mobile service robot, the accurateacquirement of current position of the robot is the prerequisite of completing thetask of navigation and tracking. WLAN and RFID are new positioning techniqueswhich have characteristics of high positioning accuracy, low power and low cost,so they are suitable for positioning of mobile robot. At present, single sensor isused to obtain information, but partial information of environmentalcharacteristics is responded due to the positioning inaccuracy of the system.Therefore, it is put forward that by using WLAN and RFID to establish thepositioning system and the information was combined using some method torealize the accurate positioning of the robot. This study is based on project of863Program “Research on Autonomous Positioning Key Technology of MobileRobot Based on WLAN/RFID Information Fusion”. Some key problems ofestablishing the WLAN and RFID system were studied in this dissertation. Thoseresearch works above lay the foundation for the development of multi-information fusion. The main contents of this dissertation are given as follows:
In the WLAN wireless positioning system, aimed at the characteristics ofthe positioning system of mobile robot, the wireless positioning technology andwireless network positioning algorithm were analyzed, and the positioningmethod was determined based on RSSI, and the path loss model was analyzed.According to the experimental environment, the signal propagation model wasdetermined and simulation experiments were done, the experimental results wereanalyzed. Using wireless router and wireless NIC the wireless positioning systemof WLAN mobile robot was established and WLAN location method wasresearched based on RSSI. Through many experiments, trilateral locating methodwas determined for WLAN locating system.
In the RFID (radio frequency identification) positioning system, aimed at the features of indoor positioning system of mobile robot such as complexenvironment and many obstacles, active radio frequency module was adapted andmobile robot location system indoors was built based on Visual C++6.0platform.In the system, four labels of the largest RSSI value were selected for positioningtabs, and based on maximum likelihood estimation method mobile error factorwas introduced, and the improved maximum likelihood estimation method wasproposed, and it was used in the system positioning system of mobile robot.
In the Federated Filtering fusion structure, UKF was adopted to filter thelocal filtering results. Aimed at the features of UKF algorithm such as the resultbeing easy to diverge and slow calculation two improved UKF algorithms wereadopted. It was proposed that the most minimum skewness strategy should becombined with fading memory. For improving the calculation accuracy theminimum skewness strategy was introduced, which decreased the samplingnumbers. And take fading memory instead of covariance matrix in the recursiveoperation to decrease the filter divergence by calculation error. The proportionalrevision factor was combined with the minimum skewness to solve the problemof non local effect. Two algorithms were proposed in order to lay a foundationfor improving the positioning accuracy and positioning speed of the fusionsystem.
On the basis of the WLAN and RFID system using single sensor could notmeet the positioning accuracy. So it was proposed that the Federated filteringtechniques was applied in the location system, and was combined with data fromWLAN and RFID to realize all accuracy improved. In the combined calculationprocess, because there was feedback reset of global filter in the local filters, theaccuracy of the local filter was also improved. It was proved from the result thatthe calculation improved was of feasibility.
Consequently it is of important theoretical significance and academic valueto establish the information fusion and location system based on WLAN andRFID, such as reduce production cost, improve the positioning efficiency,improve the quality of design, and increase the ability of market competition.
引文
[1] Z. Jia, A. Balasuriya, S. Challa. Sensor fusion-based visual target trackingfor autonomous vehicles[J]. Artificial Life and Robotics,2008,12(l-2):317-328.
[2] L. J. Devin, A. H. C. Ng, M. Sundberg, et al. Information fusion for decisionsupport in manufacturing: studies from the defense sector[J]. Theinternational journal of advanced manufacturing technology,2007,35(9-10):908-915.
[3] P. C. Lin, H. Komsuoglu, D. E. Koditsehek. Sensor data fusion for bodystate estimation in a hexapod robot with dynamical gaits[J]. IEEETransactions on Robotics,2006,22(5):932-943.
[4]张毅,罗元,郑太雄.移动机器人技术及其应用[M].北京:电子工业出版社,2007.9(1).
[5]董再励,郝颖明,朱枫.一种基于视觉的移动机器人定位系统[J].中国图像图形学报,2000,5(8):688-692.
[6]王晓晖.一种信息混合定位的移动机器人定位技术研究(硕士学位论文)[D].北京:北京航空航天大学,2007.
[7] M. Billinghurst, R. Grasset, J. Looser. Designing augmented realityinterfaces[J]. Computer Graphics(ACM),2005,39(1):17-22.
[8]刘克旺.基于红外线技术的汽车警示系统[J].装备制造,2010,(04):228,243.
[9]李昌禄,苏寒松.超声波定位系统的研究[J].实验室研究与探索,2013,(02):39-44.
[10]赵娜,李丹.浅谈蓝牙室内无线定位系统的技术实现[J].中小企业管理与科技,2012,(03):269-270。
[11]丁晓丹.超宽带无线通信关键技术研究(硕士学位论文)[D].山东:烟台大学,2007.
[12]郑航.基于Wi-Fi的室内测时差定位方法[J].移动通信,2012,(11):51-56.
[13]游战清,李苏剑,张益强等.无线射频识别技术(RFID)理论与应用[M].北京:电子工业出版社,2004:89-96.
[14]中国自动识别技术协会.条码与射频标签应用指南[M].北京:机械工业出版社,2004:29-38.
[15]何佳嘉.基于WLAN的RFID技术在电子商务应用中的研究(硕士学位论文)[D].广州:广东工业大学,2005.
[16]马宁.基于RSSI的RFID室内无线定位算法研究(硕士学位论文)[D].成都:电子科技大学,2012.
[17]李军怀,张果谋,于蕾.基于虚拟信号强度的RFID定位方法研究[J].计算机科学,2012(4):67-70.
[18] D. L. Lee, Q. Chen. A model-based WiFi localization method[C]. In:Proceedings of Infoscale,2007:1-7.
[19] University of Helsinki. Ekahau[DB/OL]. http://www.ekahau.com.
[20] S. C. Yeh, Y. Chiou, Y. J. Peng. A research for indoor LBS technology basedon RSSI prediction model[J]. Journal of Informatics&Electronics,2006,1(1):1-8.
[21] P. Castro, P. Chiu, T. Kremenek, R. Muntz. A probabilistic room locationservice for wireless networked environments[C]. In Proceedings of the3rdInternational Conference on Ubiquitous Computing (UbiComp), Georgia,USA,2001:18-34.
[22] P. Castro, R. Muntz. Managing context for smart spaces[J]. IEEE PersonalCommunications,2000,7(5):44-46.
[23] M. Youssef, A. Agrawala. Handling samples correlation in the horussystem[C]. In IEEE Infocom,2004:775-784.
[24] M. Youssef, A. Agrawala. Continuous space estimation for wlan locationdetermination systems[C]. In IEEE Thirteenth International Conference onComputer Communications and Networks,2004:161-166.
[25] M. Youssef, A. Agrawala, A. U. Shankar. WLAN location determination viaclustering and probability distributions[C]. In: Proceedings of the First IEEEInternational Conference on Pervasive Communicaitons,2003:143-150.
[26] M. Youssef, A. Agrawala. The horus location determination system[J].Wireless Networks,2008,14:357-374.
[27] A. Haeberlen, A. Rudys, A. M. Ladd, et al. Practical robust localization overlarge-scale802.11wireless networks[C]. In MobiCom’04: Proceedings ofthe10th annual international conference on Mobile computing andnetworking. New York, USA: ACM,2004:70-84.
[28]甘仲民,张更新.卫星通信技术的新发展[C],通信学报,2006,27(8):2-9.
[29] S. Dan. Real-time navigation using the Global Positioning System[J], IEEEAerospace and ElectronicSystems Magazine,1995,5:31-37.
[30] P. Bahl, V. N. Padmanabhan. RADAR: an in-building RF-based userlocation and tracking system[C]. In: Proceedings of IEEE Infocom,2000,2:775-784.
[31]尹应增.微波射频识别技术研究(博士学位论文)[D].西安:西安电子科技大学,2002.
[32] M. Lionel. LANDMAC: Indoor location sensing using active RFID[C].IEEE International Conference in Pervasive Computing andCommunications2003(IEEEPerCom2003), Dallas, TX, USA, March2004,2:701-710.
[33] A. Gad, M. Farooq. Data fusion architecture for inaritime surveillance[C].Proceeding of the fifth International Conference on Information Fusion.2002:448-455.
[34] Y. Fischer, A. Baluer. Object-oriented sensor data fusion for widemaritimesurveillance[C]. Waterside Security Conference.2010(11):3-5.
[35] D. R. Martin, C. C. Fowlkes, J.Malik. Learning to detect natural imageboundaries using local brightness, color and texture cues[J]. IEEETransaction Pattern Analysis Machine Intelligence,2004,26(5):530-549.
[36]高粱,刘晓云,廖志武.一种基于多信息融合的模糊边界检测算法[J].仪器仪表学报,2011,32(11):2506-2514.
[37]路雅宁,郭雷,李晖晖.结合边缘特征的遥感图像融合[J].光电工程,2012,39(9):18-23.
[38] M. A. A. Akhoundi. E. Valavi. Multi-sensor fuzzy data fusion using sensorswith different characteristics[J]. The CSI Journal on Computer Science andEngineering(JCSE).2010,10:1-9.
[39]赵黎明,刘贺平,张冰.多源信息融合技术及其工业应用[J].自动化仪表,2010,9:1-5.
[40]吕秀江,郑斌,于海涛.基于ARM的多传感器信息融合在工业控制中的应用[J].兵工自动化,2009,28(2):64-66.
[41] T. Krager, M. wahl. Multi-sensor integration and sensor fusion industrialmanipulation[C]. In Proc. Of the IEEE International Conference onMultisensor Fusion and Integration for Intelligent Systems,2008,8:411-418.
[42] Y. F. Zheng. Integration of multiple sensors into a robotics system and itsperformance evaluation[J]. IEEE Transactions on Robotics and Automation,1989,5:658-669.
[43] R. Chatila, M. Devy, M. Herbb. Perception system architecture forenvironment modeling and motion control of a mobile robot[C]. In2ndIAPR Workshop on Sensor Fusion and Environment Modelling,1991:830-852.
[44]钱晖.多传感器融合技术在智能机器人系统中的应用[J].上海电气技术,2010,6:44-48.
[45]咸宝金,陈松涛.智能移动机器人多传感器信息融合及应用研究[J].宇航计测技术,2010,30(2):41-44.
[46] O. Hellwich, W. C. Iedmann. Object extraction from high resolutionmultisensory image data[R]. Municlr Technical University Munich,2000:26-28.
[47]李晓峰,张树清.基于多重信息融合的高分辨率遥感影像道路信息提取[J].测绘学报,2008,37(2):178-184.
[48] O. Basir, X. H. Yuan. Engine fault diagnosis based on multi-sensorinformation fusion using dempster-shafer evidence theory[J]. InformationFusion,2007,8(4):379-386.
[49]许允之,龚乃玮,葛垚.基于信息融合方法的电机故障诊断研究[J].华北电力大学学报,2012,39(3):53-58.
[50]郭敏.多源信息融合在故障诊断技术中的应用研究[J].计算机与数字工程,2011,39(3):1-4.
[51]赵小川,罗庆生,韩宝玲.机器人多传感器信息融合研究综述[J].传感器与微系统,2008,27(8):1-4.
[52] C. H. Jiang, Z. Chen. Multisensor fusion based on decision-making inINS/CCD/SAR integrated systems[J]. Proceedings of IEEE Tencon’02,2002:695-698.
[53] K. O. Arras, N. Tomatis, R. Siegwart. Multisensor on-the-fly localizationusing laser and vision[C]. In Proceedings of the IEEE/RSJ InternationalConference on Intelligent Systems and Robots, Takamatsu, Japan,October/November,2000:462-467.
[54]沈阳自动化所研制成功新型特种机器人——基于复合机构的非结构环境移动机器人.http://www.sia.cn/xwzx/zhxw/200403/t20040315_824186.html.2004,3,15.
[55]中科院自动化所研制出移动机器人和仿生机器鱼.http://www.chinanews.com/n/2003-01-10/26/262392.html.2003,1,1.
[56]范平志,邓平,刘林.蜂窝网无线定位[M].北京:电子工业出版社,2002.
[57]李海燕,张艳.移动通信网络的移动台定位技术及应用[J].网络电信,2006,8(3):36-40.
[58]麻信洛,李晓中.无线局域网构建及应用[M].北京:国防工业出版社,2006.
[59]刘乃安.无线局域网(WLAN)——原理、技术与应用[M].西安:西安电子科技大学出版社,2004.
[60]张明华,张申生,曹健.无线局域网中基于信号强度的室内定位[J].计算机科学,2007(6):68-71,75.
[61]胡可刚,王树勋,刘立宏.移动通信中的无线定位技术[J].吉林大学学报(信息科学版),2005,23(4):378-384.
[62]黎海涛,张平.无线定位技术[J].无线通信技术,2001,4:5-8.
[63]董梅,杨曾,张健.基于信号强度的无线局域网定位技术[J].计算机应用,2004,24(12):49-52.
[64]王珂.基于WLAN的移动终端无线定位技术研究[D].郑州:郑州大学信息工程学院,2007.
[65]董梅.无线局域网定位系统的研究与实现[D].上海:华东师范大学信息科学技术学院,2004.
[66]王曼珠,曾萍,马莹.移动通信无线定位技术的发展现状[J].北京电子科技学院学报,2002,10(1):20-24.
[67] B. Azzedine, A. B. F. Horacio, Oliveira, F. Eduardo, et al. Vehicular Ad HocNetworks: a new challenge for localization based systems[J]. ComputerCommunications,2008,31(12):2838-2849.
[68]焦磊,邢建平,张军.一种非视距环境下具有鲁棒特性TOA无线传感器网络定位算法[J].传感技术学报,2007,20(7):1625-1629.
[69] M. W. Stephen, D. F. Kenneth, A. Isaac. Decentralized acoustic sourcelocalization in a distributed sensor network[J]. Applied Acoustics,2006,67(10):996-1008.
[70] J. Zheng, W. K. Kenneth, H. Lui. C. So, Accurate three-step algorithm forjoint source position and propagation speed estimation[J]. Signal Processing,2007,87:3096-3100.
[71] A. Savvides, C. C. Han, M. Srivastava. Dynamic fine-grained localization inAd-Hoc networks of sensors[C]. Proceedings of the7th Annual InternationalConference on Mobile Computing and Networking,2001:166-179.
[72] G. Q. Mao et al., Path loss exponent estimation for wireless sensor networklocalization[J]. Computer Networks,2007,51(10):2467-2483.
[73] M. Baunach, R. Kolla, C. Muhlberger. Beyond theory: development of a realworld localization application as low power WSN[C].32nd IEEEConference on Local Computer Networks,2007:872-884.
[74] K. Langendoen, N. Reijers. Distributed localization in wireless sensornetworks: a quantative comparison[J]. Computer Networks,2003,43:499-518.
[75] S. Gezici. A survey on wireless position estimation[J]. Wirless PersonalCommunications,2008,44(3):263-282.
[76]卞瑞祥,苏凯,李俊.一种无线传感器网络中的定位精度改进方法[J].计算机应用与软件,2009,26(6):154-159.
[77]李晓莺,曾启铭. NDIS网络驱动程序的研究与实现[J].计算机应用,2002,22(4):60-61.
[78] A. D. Koutsoul, F. Secol, A. R. Jimenez1. Preliminary localization resultswith an RFID based indoor guiding system[J]. Intelligent Signal Processing,2007:1-6.
[79] D. Simon, H. E. Sherief. Real-time navigation using the global positioningsystem[J], IEEE Aerospace and ElectronicSystems Magazine,1995,10(1):31-37.
[80]周晓光,王小华.射频识别(RFID)技术原理与应用实例[M].人民邮电出版社,2006,12.
[81]杨殿阁,连小珉,张涛.基于北斗卫星的车辆组合导航系统开发[J].清华大学学报(自然科学版),2008,48(5):856-859.
[82]孙瑜.射频识别(RFID)室内定位算法研究(硕士学位论文)[D].成都:西南交通大学,2005.
[83]王殿君,任福君,兰云峰.基于有源RFID信号强度的室内移动机器人测距方法[J].科技导报,2009,27(15):76-79.
[84]陈永光,李修和.基于信号强度的室内定位技术[J].电子学报,2004,32(9):1456-1458.
[85]罗辉.四探针智能测试仪的研发(硕士论文)[D].杭州:电子科技大学,2010,1,1.
[86] K. Tateishi, T. Ikegami. Estimation method of attenuation constant duringlocalization in RSSI[C]. Proceedings of International Symposium onCommunications and Information Technologies, Vientiane, Lao PDR,2008:482-487.
[87] Y. Lu, C. Lai, C. Hu, et al. Path loss exponent estimation for indoor wirelesssensor positioning[J]. KSII Transactions on Internet and Infromation System,2010,4(3):243-257.
[88] H. Ahn, W. Yu, Environmental-adaptive RSSI-based indoor localization[J].IEEE Transactions on Automation Science and Engineering,2009,6(4):626-633.
[89] I. Yamada, T. Ohtsuki. An indoor location estimation method by maximumlikelihood using RSSI[C]. Proceedings of3rh Sensor Network Conference,Tokyo, Japan,2006:37-41.
[90] R. Zemek, S. Hara, K. Yanagihara, et al. A joint estimation of target locationand channel model parameters in an IEEE802.15.4-based wireless sensornetwork[C]. Proceedings of18th Annual IEEE International Symposium onPersonal, Indoor and Mobile Radio Communications, Athens, Greece,2007:1-5.
[91] R. Zemek, D. Anzai, S. Hara, et al. RSSI-based localization without a priorknowledge of channel model parameters[J]. International journal of wirelessinformation networks,2008,15:128-136.
[92]卡尔曼滤波. http://blog.sina.com.cn/s/blog_4b01c72b0100sp0b.html.2012,10,16.
[93]杨少委,张旭东.基于SR-UKF的高动态GPS信号参数估计[J].通信技术.2010,43(09):9-11.
[94]刘丽雯,张崇巍,徐玉华等.基于UKF的室内移动机器人定位问题[J].合肥工业大学学报(自然科学版),2009:19-13.
[95]潘泉,杨峰,叶亮,等.一类非线性滤波器——UKF综述[J].控制与决策,2005,5:481-489,494.
[96] S. J. Julier, J. K. Uhlmann. Unscented filtering and nonlinear estimation[C].Proceedings of the IEEE,2004,92(3):401-422.
[97]李丹,刘建业,熊智,等.基于最小偏度采样的卫星自主导航SRUKF算法[J].南京航空航天大学学报,2009,2:54-58.
[98]姜雪原,马广富,胡庆雷.改进型UKF滤波算法的卫星姿态估计[J].哈尔滨工程大学学报,2005,8:544-549.
[99]刘海洋,杨乐,陈杰.改进无色卡尔曼滤波算法在GPS系统中的应用[J].系统仿真学报,2009,8:4859-4865.
[100]罗涌,钟洪声.一种改进的UKF算法在目标跟踪系统建模中的应用[J].遥测遥控,2008,1:55-58.
[101]章飞,周杏鹏,陈小惠.基于衰减记忆滤波的平方根UKF被动目标跟踪算法[J].测控技术,2010,4:22-26.
[102] S. J. Julier. The spherical simplex unscented transformation[C]. Proceedingof the American Control Conference. Denver Colorado, IEEE,2003,3:2430-2434.
[103] M. N. Andersen, K. Wheeler. Filtering in hybrid dynamic bayesiannetworks[C]. IEEE International Conference on Acoustics, Speech, andSignal Precessing,2004:773-776.
[104] N. Tomatis, Nourbakbsh, R. Siegwart. Combining topological and metric:a natural integration for simultaneous localization and map building[C].Proceedings of the Fourth European Workshop on Advanced MobileRobots, Sweden,2001:21-29.
[105]郭剑辉,赵春霞.尺度Unscented变换在同时定位与地图创建算法中的应用研究[J].兵工学报,2008,29(7):859-864.
[106]秦永元,张洪钺.卡尔曼滤波与组合导航原理[M].西北工业大学出版社,2010,3.
[107] S. J. Julier, J. K. Uhlmann. Reduced sigma point filters for the propagationof means and covariances through nonlinear transformations[C].Proceedings of American Control Conferece,2002,2:887-892.
[108] S. J. Julier. The scaled unscented transformation[C]. Proceedings of theAmerican Control Conference, Anchorage, AK,2002:4555-4559.
[109]李雪鹏,张幼群,包括. UKF的改进算法及其在伪卫星定位中的应用[J].测绘科学技术学报,2008,4(2):108-111.
[110]刁鹏,谢聂,吴训忠.低成本MINS/GPS组合导航系统的设计实现[J],传感技术学报,2009,9(9):1366-1370.
[111]韩崇昭,朱红艳,段战胜.多源信息融合[M].清华大学出版社,2006,4.
[112]罗元,邵帅,张毅.基于信息融合的移动机器人定位与路径规划[J].计算机应用,2010,11:3091-3096.
[113]陈增强,袁著祉,张燕.基于神经网络的非线性预测控制综述[J].控制工程,2002,9(4):35-38.
[114]陈少斌,蒋静坪.运用异质传感器信息融合的移动机器人自定位[J].控制理论与应用,2008,25(5):883-886.
[115]刘洞波,刘国荣,喻妙华.融合异质传感信息的机器人粒子滤波定位方法[J].电子测量与仪器学报,2011,1:38-43.
[116]原新,王东阳,严勇杰.基于推算定位和超声波定位融合的机器人自主定位技术[J].海军工程大学学报.2009,21(5):67-72.
[117]杨清梅,王立权,王岚.基于数据融合的拱泥机器人定位方法[J].哈尔滨工程大学学报.2003,4:363-367.
[118] N. A. Darlson, M. P. Berauducel. Federated kalman filter simulationresults[J]. Navigation,1994,41(3):297-321.