摘要
本文着重讨论图履带式能力风暴越野版移动机器人AS-RF实现自主爬楼梯所需要的感知识别楼梯尺寸和机器人姿态的方法,并基于多传感器融合自主爬楼梯的控制,讨论机器人爬楼梯的控制方法和控制系统结构。
通过对移动机器人(AS—RF)传感器系统以及扩展的传感器系统的分析,本文的主要内容有:
首先研究了移动机器人基于各传感器的姿态感知,主要研究利用声纳、激光感知楼梯环境,得到两个控制参数;将单目CCD得到的图像进行边缘提取、线段拟合和台阶判别处理,方便后面多传感器感知数据融合。
然后介绍多传感器融合理论并针对D-S证据理论存在的问题采用了一种改进的D-S证据理论方法。本文根据传感器的局部决策先算出局部决策值,构造整个系统的支持矩阵,然后求这个支持矩阵的特征向量,以此作为各个传感器的可信度。这个可信度就可以作为各个传感器的权值,以此修正D-S证据的融合算法。最后,给出了一个对比仿真实验,证明此算法在处理错误或大偏差证据的有效性。
最后通过多传感器融合后的感知参数来控制移动机器人自主爬楼梯控制,自主爬楼梯控制可以分成单层楼梯的自主爬行和多层楼梯的自主爬行。在单层自主控制的基础上,通过机器人自动检测是否到达楼梯顶部平台、自动寻找下一层楼梯即可实现多层楼梯的自主爬行,控制机器人在爬楼梯过程中自主调节位姿以保证安全爬上楼梯。
This article researches on the method that can be used by AS-RF mobile robot in climbing the stairs and apperceiving the position of stairs and robot. Based on multi-sensors amalgamation, this article also talks about the configuration of control system and its control policy.
After making analysis to the AS-RF system and the sensors' extended system, the main content of this article can be divided as follows:
First of all, apperceiving pose and position of robot based on several sensors, such as sonar and laser, is researched in order to achieve the control parameters; the images came from the CCD are disposed by edge distilling, line fitting and sidestep distinguishing in order to make facility to the following data fusion of multi-sensors.
Secondly the theory of multi-sensor fusion is introduced and an improved method of D-S evidential reasoning is adopted to slove the present problem of D-S evidential reasoning. The paper calculates the local decision on the basis of the measurement, on which the support matrix is based.Then the eigenvector is gained from the support matrix, and it is the reliability vector of the system. The paper improves the D-S evidential reasoning by giving a weight to evidence that equals to the evidence's reliability. At last, an example is given out.It proves that the method described above is valid when some evidences are false or far from the real value.
Finally, the robot is controlled to climb stairs through apperceiving parameters, the policy can be divided into two parts: monolayer stair climbing and multi-stairs climbing. The multi-stairs climbing can be accomplished through checking whether AS-RF has been in the roof of stair and finding the next floor's stair based on monolayer stair climbing. The pose of robot should be controlled independently by itself in order to guarantee the safety in climbing stairs.
引文
[1]蔡自兴.机器人学(M),北京:清华大学出版社,2000.
[2](美)克来格,机器人学导论[M],北京,机械工业出版社,2006.
[3]杨淑莹,图像模式识别[M],北京,清华大学出版社,2005.
[4]李弼程,彭天强,彭波,智能图像处理技术[M],北京,电子工业出版社,2004.
[5]孙全,叶秀清,顾伟康,一种新的基于证据理论的合成公式[J].电子学报,2000:28(8):117-119.
[6]Mallat Stephane,Wen Liang Hwang,Singularity Detection and Processing with Wavelets[J],IEEE Trans.On Information Theory,1992,38(20):617-643.
[7]J.A.Perez,J.A.Castellanos,J.M.M.Montievl,J.Neira,J.D.Tardos,Continuous mobile robot localization:vision vs.laser[J],IEEE International Conf.on Robotics and Automation,Michigan,USA,1999:2917-2923.
[8]SaschaA.Stoeter,PaulE.Rybski,MariaGini,NikolaosPapanikolopoulos Autonomous Stair-Hopping with Scout Robots:inti.Conference on intelligent Robots and Systems EPFL,Lausanne,Switzerland,2002:721-725.
[9]Kazuo Sugibara and John Smith.Genetic Algorithms for Adaptive Motion Planning of an Autonomous Mobile Robots[A].Problems IEEE Trans.SMC[C],M.I.T.,U.S,1997,PP:138-143.
[10]杨万海,多传感器数据融合及其应用,西安,西安电子科技大学出版社,2004.
[11]蒋杰,新型智能车辆视觉系统及图像处理技术的研究[D].长春:吉林大学硕士学位论文,2003.
[12]Xu L.,Oja E.,A new curve detection method:Randomized Hough transform(RHT),Pattern Recognition Letters,1990,11(5):331-338.
[13]Delanoy,R.L.,"Toolkit for Image Mining:User-Trainable Search Tools," Lincoln Laboratory Journal,8(2),p p.145-160,1995.
[18]Streilein,W.,Waxman,A.,Ross,Liu,F.,Braun,M,Fay D.,Harmon,P.,and Read C.H.,(2000),FusedMulti-Sensor Image Mining for Feature Foundation Data,To appear in Proceedings of Third InternationalConference on Information Fusion,Pans,July 2000.
[14]Carpenter,G.A.Carpenter & Grossberg,S.Markuzon,J.H.,Reynolds,J.H.&Rosen,D.B.,FuzzyARTMAP:A neural architecture for incremental supervised learning of analog multidimensional maps.IEEE Transactions on Neural Networks.3(S).698-713,1688-1692.
[15]Aguilar,M.,Fay D.A.,Ireland D.B.,Racamato J.P.,Ross W.D.,and Waxman A.M.,"Field Evaluations of Dual-Band Fusion for Color Night Vision," Proc.of SPIE Cinf. On Enhanced and Synthetic Vision 1999,SPIE-3691,pp.168-175.
[16]朱志刚等.数字图像处理[M1.北京:电子工业出版社,1998.
[17]乔凤斌,杨汝清.六轮移动机器人爬楼梯能力分析[J].机器人,2004,26(4):301-305.
[18]王壮,胡卫东,郁文贤等.数据融合中的Dempster-Shafer证据理论[J].火力与指挥控制,2001:26(3):6-10.
[19]任从富,耿卫东,潘云鹤.面向数据融合的DS方法综述[J].电子学报,2001:29(3):393-396.
[20]孟伟,洪炳镕,韩学东,基于D-S证据理论的月球探测机器人的信息融合.哈尔滨工业大学学报,2003:1041-1042.
[21]潘巍,王阳生,杨宏戟,D-S证据理论决策规则分析.计算机工程与应用,2004:15-17.
[22]王巍,夏玉华,梁斌等.月球漫游车关键技术初探[J].机器人,2001,23(3):280-284.
[23]袁军,黄心汉,陈棉江,基于多传感器的移动机器人信息融合、控制结构和应用,机器人,1997,16(5),p.p.313-320.
[24]张明路、孟庆浩等,基于多传感器的移动机器人对感知环境的识别,天津大学学报,1998,31(4):p.p47-51.
[25]梁华,移动机器人路径规划与视觉系统基础研究[D].杭州:浙江大学硕士学位论文,2002.
[26]王珂,郭显久,庄严,王伟,基于单目视觉的Robocup全自主移动机器人定位研究[J].第五届全球智能控制与自动化大会.中国杭州.2004:4797-4801.
[27]肖本贤,余炎峰,余雷,陈昊,基于免疫遗传算法的移动机器人全局路径规划[J].计算机工程与应用,2007,43(30):91-93.
[28]张小虎,李由,李立春,王鲲鹏,于起峰,一种基于梯度方向直方图的直线轮廓提取新方法[J].光学技术,2002,32(6):50-53.
[29]邬永革,杨静宇,多传感器数据融合的概念、方法和实现,机器人,1995:Vol.17,增刊,p.p.564-568.
[30]ZezhongXu,Jilin Liu,Zhiyu Xiang,Han Li,Map building for indoor environment with laser ranger scanner[J],IEEE 5th Int.Conf.on Intelligent Transportation System,2002:136-140.
[31]Borges G.A,Aldon M.J,Gil T,An optimal pose estimator for map-based Mobile robot dynamic localization:experimental comparison with the EKF[J],IEEE International Conf.on Robotics and Automation,2001:1585-1590.
[32]沈庭芝,方子文.数字图像处理及模式识别(M).北京:北京理工大学出版社 1998.
[33]陈荣保,杨峰,李天庆.一种基于单目视觉的机器人监测方法研究.仪器仪表 学报增刊.合肥.2007:585-589.
[34]蒋杰,新型智能车辆视觉系统及图像处理技术的研究[D].长春:吉林大学硕士学位论文,2003.
[35]汪哲慎,李翠华,基于改进Hough变换的建筑目标搜索与识别[J].中国图像图形学报,2005.10(4):463-467.
[36]Deng Y,Shi W K,Zhu Z F.Efficient combination approach of conflict evidence[J].Journal of Infrared and Millimeter Waves,2004:23(1):27-32.
[37]Murphy C K.Combining belief functions when evidence conflicts[J],Decision Support Systems,2000:29(1):1-9.
[38]Lefevre E.Belief function combination and conflict management[J].Information Fusion,2002:(3):111-114.
[39]Dubios H,Representation and combination of uncertainty with belief functions and possibility measures[J].Computayional Intelligence,1998(1):88-101.
[40]Walley P.Measures of Uncertainty in Expert System[J],Artif Intell,1996:18(3):1-58.
[41]Yager R R.On the aggregation of prioritized belief structure[J].IEEE Trans on Syst,Man and Cybern,1996,26(6):708-719.
[42]Wu Y G,Yang J Y,Lin K,et al.On the Evidence Inference Theory[J].Information Science,1996:89(3/4):245-260.
[43]高国富,谢少荣,机器人传感器及其应用[M],北京,化学工业出版社,2005.
[44]徐从富,耿卫东,潘云鹤,面向数据融合的DS方法综述[J].电子学报,2001:29(3):393-396.
[45]孟伟,洪炳镕,韩学东,基于D-S证据理论的月球探测机器人的信息融合.哈尔滨工业大学学报,2003.35(9):1041.
[46]许丽佳,D-S理论在信息融合中的改进.系统工程与电子技术.2004.26(6):718-719.
[47]陈一累,王俊杰,一种D-S证据推理的改进方法.系统仿真学报,2004.16(1):29.
[48]陈天璐,阙沛文,D-S证据理论在数据融合中失效问题的研究*.传感器技术,2004.23(12):25-27.