文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > IT/计算机 > Accurate Mobile Robot Dead-Reckoning with a
侵权投诉

Accurate Mobile Robot Dead-Reckoning with a

基于DR航迹推算卡尔曼滤波的IEEE论文

ShortPapers_______________________________________________________________________________

AccurateMobileRobotDead-ReckoningwithaPrecision-CalibratedFiber-OpticGyroscope

HakyoungChung,LauroOjeda,andJohannBorenstein

Abstract—Thispaperdescribestwomethodsaimedatimprovingdead-reckoningaccuracywithfiber-opticgyroscopes(FOGs)inmobilerobots.ThefirstmethodisaprecisioncalibrationprocedureforFOGs,whichef-fectivelyreducestheilleffectsofnonlinearityofthescale-factorandtem-peraturedependency.Thesecondmethodistheimplementationofanindi-rectfeedbackKalmanfilterthatfusesthesensordatafromtheFOGwiththeodometrysystemofthemobilerobot.

Thepaperalsoprovidesexperimentalresultsandcomparestherelativeeffectivenessofthetwomethodsasimplementedonafour-wheeldrive/skid-steerPioneerATmobilerobot.

IndexTerms—Dead-reckoning,fiber-opticgyroscope,indirectfeedbackKalmanfilter,mobilerobot,odometry.

TABLEI

TECHNICALSPECIFICATIONSFORTHEE-CORE

RD2100FOG.(COURTESYOF

Accurate Mobile Robot Dead-Reckoning with a

[KVH])

I.INTRODUCTION

Untiljustafewyearsago,allbutthehighest-grademechanicalgyro-scopeshadlargebiasdriftrates.However,thissituationchangedwiththereductionincostofcommerciallyavailablefiber-opticgyroscopes(FOGs).Forexample,theKVHE-CoreRD2100gyro1[1],whichwasthemodelusedinourexperiments,costsabout$2,000anditsdriftrateisspecifiedas(atmost)0.002 /s=0.12 /min=7.2 /hr(seeTableIforspecificationsaspublishedbythemanufacturer).However,withthereductionofbiasdrifterrorsothermeasurementerrorsgaininrel-ativesignificance.Mostnotableamongthesehithertosecondaryerrorare:1)thenonlinearityofthescalefactorand2)thesusceptibilityofthesensortochangesinambienttemperature.ThesetwoerrorscanbereducedsubstantiallybyperforminganindividualcalibrationforeachFOG[2].

Whenmorethanonesensormodalityisusedtomeasureoneparam-eter,suchastherelativeorientationofamobilerobot,thentheques-tionariseshowtofusethereadingsfrombothsensormodalities.Bar-shanandDurrant-Whyte[3]proposedadirectKalmanfilterthates-timatestherobotpositionandattitude.Otherresearch[4],[5]utilizesgyroscopesandencoderstoobtainacloseestimateoftherobotposi-tion.However,thesemethodsneedfastcomputationsduetotherapidlyvaryingnatureoftherobotdynamics.Inthispaper,weproposeanin-directKalmanfilterthatestimateserrorstatesanddoesnotrequirefastcomputation[6].

ThecalibrationprocedurefortheFOGisdescribedinSectionII,whilethedesignofourKalmanfilterisdetailedinSectionIII.Thecompletedead-reckoningsystemwasimplementedonaPioneerATfour-wheeldrive/skid-steermobileplatform,andexperimentalresultsareprovidedinSectionIV.

II.GYROSCOPECALIBRATION

Onewell-knownsourceoferrorsingyroscopesisthestaticbiasdrift.Inthispaper,wearenotfocusingonthis(significant)errorsourcebe-causeatrivialshort-termcalibrationmethodiseffectiveindealingwiththisproblem:priortoeachmobilerobotmission(i.e.,whiletherobotisstandingstill),thegyrooutputissampledfor,say,5–10s,andthereadingsareaveraged.Then,oncetherobotismoving,thisaveragedstaticbiasvalueissubtractedfromallsubsequentgyroreadings.Be-causeoftherelativelylowstaticbiasdriftinFOGs,missiondurationsontheorderof10minarefeasiblebeforeanewstaticbiasdriftvalueshouldbedeterminedbyrepeatingtheaboveshort-termcalibration.Oncethestaticbiaserrorisreducedintheabovedescribedway,twoothersourcesoferrorsbecomedominant:1)thenonlinearityinthescalefactorand2)thegyro’ssusceptibilitytochangesinambienttemperature.Thissectionbrieflyexplainsthemathematicalmodelfortheseerrors,whileamoredetailedtreatmentcanbefoundin[2].A.NonlinearityoftheScaleFactor

Thebasicideaformeasuringscalefactorerrorsistorotatethegyroatavaryingbutpreciselycontrolledinputrates!whilecomparingthemeasurementoutputofthegyro!gtotheknowninputrates.Thedif-ferencebetweenbothvaluesistheerrorintroducedbythenonlinearityinthescalefactor".

"=!g0!:

ManuscriptreceivedMarch27,2000;revisedOctober6,2000.ThispaperwasrecommendedforpublicationbyAssociateEditorG.HagerandEditorA.DeLucauponevaluationofthereviewers’comments.ThisworkwassupportedbyDARPAunderAwardDAAE07-98-C-L029,DOEunderAwardDE-FG04-86NE37969,andSeoulNationalUniversityofTechnology.Thispaperwaspresentedinpartatthe2000IEEEInternationalConferenceonRoboticsandAutomation,SanFrancisco,CA,April24–28,2000.

H.ChungiswiththeDepartmentofControlandInstrumentationEngi-neering,SeoulNationalUniversityofTechnology,Seoul139-743,Korea(e-mail:chy@duck.snut.ac.kr).

L.OjedaandJ.BorensteinarewiththeAdvancedTechnologiesLabo-ratory,UniversityofMichigan,AnnArbor,MI48109-2110USA(e-mail:lojeda@umich.edu;johannb@umich.edu).

PublisherItemIdentifierS1042-296x(01)02770-8.

1KVH,

TinleyParkIL60477USA.Availableatdoc.guandang.net

1042–296X/01$10.00©2001IEEE

(1)

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email