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 > 工程科技 > 交通运输 > Adaptive Regulation of Amplitude Limited Robot Manipulators With Uncertain Kinematics and Dynamics
侵权投诉

Adaptive Regulation of Amplitude Limited Robot Manipulators With Uncertain Kinematics and Dynamics

不确定运动学、动力学影响下输入受限自适应正定系统设计

488IEEETRANSACTIONSONAUTOMATICCONTROL,VOL.52,NO.3,MARCH2007

AdaptiveRegulationofAmplitudeLimitedRobotManipulatorsWithUncertainKinematicsandDynamics

W.E.Dixon

Abstract—Commonassumptionsinmostofthepreviousrobotcon-trollersarethattherobotkinematicsandmanipulatorJacobianareperfectlyknownandthattherobotactuatorsareabletogeneratethenecessaryleveloftorqueinputs.Inthisnote,anamplitude-limitedtorqueinputcontrollerisdevelopedforrevoluterobotmanipulatorswithun-certaintyinthekinematicanddynamicmodels.Theadaptivecontrolleryieldssemiglobalasymptoticregulationofthetask-spacesetpointerror.Theadvantagesoftheproposedcontrollerincludetheabilitytoactivelycompensateforunknownparametriceffectsinthedynamicandkinematicmodelandtheabilitytoensureactuatorconstraintsarenotbreachedbycalculatingthemaximumrequiredtorqueapriori.

I.INTRODUCTION

Foraroboticsystemtointeractwithandexecutetasksinthetask-space,atransformationbetweenthepositionandorientationoftherobotwithrespecttoobjectslocatedinthetask-spaceistypicallyre-quired.Sincetherobotiscontrolledthroughinputstothelinkactua-tors,therobotkinematicsandmanipulatorJacobianareusedtorelateatask-spacecoordinatesystemtocoordinatesystemsattachedtoeachactuator(i.e.,thejoint-space).AcommonassumptioninmostofthepreviousrobotcontrolliteratureisthatthekinematicsandmanipulatorJacobianareperfectlyknown.Thisassumptionlimitsrobustnessbe-causeuncertaintyintherobotkinematicsandJacobianmayleadtodegradedperformanceorunpredictableresponses.

Uncertaintyinthekinematicmodelofarobotisapracticalproblem.Theend-effectorofamanipulatorisofteninterchangedwithotherendeffectortoolswithdifferentlengths;themanipulatorpicksupanob-jectwithanunknownlengthwiththeintentionofusingitasatool;orgenericrobotdatasheetsmaybeusedthatarenotspeci ctothepartic-ularrobot.Forexample,thePuma560haswellpublishedkinematicsbutmanydifferentversionsoftherobotareonthemarketthathavethesamekinematicstructurebutdifferentkinematicparameters(e.g.,duetodifferencesinthemanufacturers).Fromtheseexamples,itisalsoclearthatuncertaintyinthekinematicmodelisaseparateproblemfromuncertaintyinthedynamicmodel.Intheinvitedpaper[1],Ari-motodescribestheimportanceoftheproblemwithuncertainkinematicparametersandstatesthatresearchwhichtargetsthisproblemisatthebeginningstages.

TheobservationsbyArimotoaresupportedbyareviewofliteraturewhichyieldsrelativelyfewdifferentapproachestoaddressthistopic.In[28],ShahamiriandJagersandusedanullspace-biasedNewton-stepvi-sualservostrategywithaBroydentypeJacobianestimationforon-linesingularitydetectionandavoidanceinanuncalibratedvisualservocon-trolproblem.Basedonasimilaridea,Piepmeieretal.publishedase-ManuscriptreceivedAugust5,2005;revisedApril12,2006andAugust31,2006.RecommendedbyAssociateEditorA.Astol .ThisworkwassupportedinpartbytheNationalScienceFoundationunderCAREERAwardCMS-0547448,bytheAirForceOf ceofScienti cResearchunderContractsF49620-03-1-0381andF49620-03-1-0170,bytheAFRLunderContractFA4819-05-D-0011,andbyResearchGrantUS-3715-05fromBARD,theUnitedStates–IsraelBinationalAgriculturalResearchandDevelopmentFundattheUniversityofFlorida.

TheauthoriswiththeMechanicalandAerospaceEngineeringDepart-ment,theUniversityofFlorida,Gainesville,FL32611-6250USA(e-mail:wdixon@u .edu).

Colorversionsofoneormoreofthe guresinthispaperareavailableonlineatdoc.guandang.net.

DigitalObjectIdenti er10.1109/TAC.2006.890321

riesofpapers(e.g.,see[21]–[25])thatexploitdynamicquasi-Newtonrecursiveleast-squaresestimatorstosolveavarietyofvisualservocon-trolproblemsdespiteanuncertainkinematicmodelanduncertaintyinthecameramodel.

SomeresultshavebeenpublishedbyCheahetal.in[4]–[6]thatad-dressuncertaintyinthekinematicmodel.Speci cally,Cheahetal.de-velopedseveralapproximateJacobianfeedbackcontrollersin[4]–[6]thatexploitastatic,best-guessestimateofthemanipulatorJacobiantoachievetask-spaceregulationobjectivesdespiteparametricuncer-taintyinthemanipulatorJacobian.Asreportedin[3],adrawbackofthecontrollersin[4]–[6]isthatthetask-spacevelocityoftherobotend-effectorisrequiredtobemeasurable,andthecontrollerin[5]re-quiresthecomputationofanestimatefortheJacobianinverse.In[3],Cheahetal.resolvetheseissuesbydevelopingacontrollerthatexploitsastatic,best-guessestimateofthemanipulatorJacobiantoachieveatask-spaceregulationresult.

Controllersthatexploitcombinationsofproportional,integral,orderivativefeedbackterms(i.e.,PDandPID)suchastheseminalworkin[19]canbeusedtoachievejoint-spaceregulationinthepresenceofuncertaintyinthedynamicmodel;however,task-spaceregulationhasyettobeprovenbysuchacontrollerwhenthekinematicsandma-nipulatorJacobianarecorruptedbyuncertainty.APIDcontrollersuchas[19]couldbeusedtoachievejoint-spaceregulationthatiscoupledwithanadaptivepathplannerthatcompensatesforthekinematicuncer-tainty.Thisisavalidapproachthatisalsoatthebeginningstagesofre-searchwithsomeresultspresentedin[7]and[14];however,shiftingtheabilitytocompensateforthekinematicuncertaintyonthepathplannermaynotbereasonableforsomeapplications.Thatis,effortsshouldalsobedevelopedthatcompensatefortheuncertaintythroughthecon-trollersothatagenericpathplannercanbeeasilyincorporated.

Thecontrollerdevelopedinthisnote(andthepreliminaryworkin[12])isthe rstadaptiveresultfortheregulationproblemunderthepresenceofkinematicanddynamicuncertainty.Asintherecentre-sultin[3],thecontrollerdoesnotrequirethetask-spacevelocityoftherobotend-effectortobemeasurable,doesnotrequiretheinverseoftheestimatedJacobiantobecomputed,anddoesnotrequireexactmodelknowledgeoftherobotdynamics.Thecurrentresultactivelycompen-satesforuncertaintyinthegravityandstaticfrictioneffects.Incontrasttotheuseofastatic,best-guessestimateasin[3],thecontrollerinthisnotealsoexploitsafeedforwardcontroltermthatactivelycompensatesfortheparametricuncertaintyintheJacobian.Theadaptivecontrollerinthisnoteprovidesthe rstresultthateliminatestheassumption(re-quiredbythepreviousrobustcontrollers)thatthemismatchbetweentheJacobianandtheJacobianestimatebeboundedasfollows:

^(q)k kJ(q)0J

(1)

where issomeknownpositiveconstant.Moreover,theconstraints

onthefeedbackgainsthatareimposedby(1)(i.e.,high-gainrobustfeedback)areeliminatedbytheproposedadaptivecontroller.Thede-velopmentoftheadaptivecontrollerrequiresaninnovativearrange-mentoftheuncertaintyintheopen-looperrorsystemthatisenabledbyusingappropriatesaturationmethodstoboundfunctionsinaspe-ci cmanner.StrategicuseofthesaturationfunctionscoupledwithaninnovativenonquadraticLyapunovfunctionalsoallowsthecontrolef-forttobeaprioribounded.

Thefactthatthecontrollercanbeaprioriboundedisasigni cantaddedadvantage.Thepracticalimplicationsarethattheactuatorscanbeappropriatelysized(orthecontrolgainssettospeci cthresholds)withoutrequiringanadhocsaturationscheme(thatwouldnotbein-cludedinthestabilityanalysis)toprotecttheactuator.Thecommonassumptionthattherobotactuatorsareabletogeneratethenecessary

0018-9286/$25.00©2007IEEE

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