移动机器人导航使用模糊极限周期在cluttered环境(IJISA-V6-N7-2)

I.J.IntelligentSystemsandApplications,2014,07,12-21PublishedOnlineJune2014inMECS()DOI:10.5815/ijisa.2014.07.02Copyright©2014MECSI.J.IntelligentSystemsandApplications,2014,07,12-21MobileRobotNavigationusingFuzzyLimit-CyclesinClutteredEnvironmentFatmaBoufera*,***MathematicsandComputerScienceDepartment,UniversityofMascara,Mascara,Algeria**LaboratoryofResearchesinIndustrialComputing&Networks,FacultyofExactandAppliedSciences,UniversityofOran,AlgeriaE-mail:[e-mail]*FatimaDebbatE-mail:[e-mail]ascal,UBP-UMRCNRS6602,ClermontFerrandFranceE-mail:[e-mail]**MohamedFaycalKhelfiE-mail:[e-mail]—Thispaperproposesahybridapproachbasedonlimit-cyclesmethodandfuzzylogiccontrollerfortheproblemofobstacleavoidanceofmobilerobotsinunknownenvironment.Thepurposeofhybridizationconsistsontheimprovementofbasiclimit-cyclemethodinordertoobtainsafeandflexiblenavigation.Theproposedalgorithmhasbeensuccessfullytestedindifferentconfigurationsonsimulation.IndexTerms—MobileRobot,ObstacleAvoidance,Limit-CyclesMethod,FuzzyLogicI.INTRODUCTIONInunstructuredenvironments,robotcontrollermustbeabletooperateunderconditionsofimprecision,uncertaintyandrespondefficientlytoevents[1],[2],[3],[4],[5].Theobstacleavoidanceisanessentialcomponenttoachievesuccessfulnavigation.Severaltrajectorytrackingandpathfollowingalgorithmshavebeenproposedtosteerthemobilerobotalongapathtoadesiredgoal.Manyresearchesturnedtheirattentiontotheobstacleavoidanceproblem.Theydevelopedinterestingreal-timemethodsandalgorithmsformobilerobotsnavigationinunknownenvironments.Somefamousclassicalobstacleavoidancemethodsthatmustbecitedare:-Thepotentialfieldmethodintroducedby[6]andimprovedin[7].Thismethodassumesthattherobotisdrivenbyvirtualforcesthatattractittowardsthegoal,orrejectitawayfromthetowardsthegoal.-Vectorfieldhistogrammethodintroducedby[8]andimprovedby[9].Itcorrespondstolocaloccupancygrid,constructedfromthesensorsoftherobot.-Thedeformablevirtualzonemethod,introducedby[10].Itcanbeappliedtoanyformofobstacle,henceitsefficiency.-Limit-cyclesmethodintroducedby[11]andimprovedby[12](cf.SectionIII).Otherinterestingobstacleavoidancealgorithmsinspiredfromartificialintelligencetoolsandrelativelysuitableforreal-timeandembeddedapplicationscanbecited[13]:-NeuralNetwork:Thisapproachisappliedin[14]todeterminetheoptimalneuralnetworksstructureforreal-timeobstacleavoidancetask.Theauthorin[15]proposedaneuralnetworkthatusesQlearningreinforcementtechniqueforsolvingtheproblemofobstacleavoidance.-Multi-AgentSystem:Theauthorin[16]proposedanewlocalcollisionavoidancealgorithmbetweenmultiplerobotsforreal-timenavigationispresented.Thisalgorithmisbasedonmulti-agentsystemandquadraticoptimizationmethodforacollisionfreenavigationandrobotpathplanning.-HybridgeneticAlgorithm:Theauthorsin[17]usedahybridgeneticalgorithmwithneuralnetworkandlocalsearchmethodforsolvingtheproblemoffindingtheoptimalcollisionfreepathincomplexenvironmentsformobilerobot.-ParticleSwarm:Optimization(DPSO)tosolvetheproblemofobstacleavoidanceformulti-robotsystem.Thispaperdealstheobstacleavoidanceproblemforanon-holonomicmobilerobotinunstructuredenvironmentwhileusinganewhybridapproachbasedonlimit-cyclesmethodandfuzzylogicfordifferentshapesofobstacles.Thelimitcyclesmethoddespitetoitseffectivenessinobstacleavoidancehassomeimperfectionspresentedbyoscillationsintherobottrajectoryduringitspassagearoundtheobstacleespeciallywhenitisveryclose[12].MobileRobotNavigationusingFuzzyLimit-CyclesinClutteredEnvironment13Copyright©2014MECSI.J.IntelligentSystemsandApplications,2014,07,12-21Tosolvethisproblemandwhileinsuringthattherobotpathwillbemoreflexible,thispaperdevelopsaninferencesystembasedonfuzzylogicaddtotheobstacleavoidancemodule.Besidesthisintroduction,thestructureofthepaperisasfollows:SectionIIgivesthespecificationoftherobotposture.SectionIIIpresentsthecontrolarchitecturebasedonthelimit-cyclesmethodandinferencefuzzysystem.SectionIVisdevotedtothedescriptionandanalysisofthesimulationresults.ConclusionsandfutureworkaregiveninSectionV.II.ROBOTPOSTURELetx(t),y(t)andz(t)representtherobot’spositionattimetinsomeglobalcoordinatesystem(Ω),andlettherobot’sorientation(headingdirection)berepresentedbyθ(t).Thequadruplet(x,y,z,θ)describestherobot’skinematicconfiguration.Inourcase,theusedrobot(Fig.1)hastwoindependentlydrivewheelsatadistanced,withfixedmaximumwheelvelocityVmaxandmaximumwheelaccelerationamax..Fig.1.RobotconfigurationinaCartesianreferenceframe(̇̇̇)()()With::Configurationstateoftheunicycleatthepointofabscissaandordinateaccordingtothemobilereferenceframe.:Linearvelocityoftherobotatthepoint.:Angularvelocityoftherobotatthepoint.III.PROPOSEDHYBRIDOBSTACLEAVOIDANCEThissectiondescribesthebehaviorcontrolarchitecturefornavigation,obstacleavoidanceandattractiontothetargetbasedonlimit-cyclesmethod[12]andfuzzylogicinunstructuredenvironment.Fig.2.Controlarchitectureformobilerobotnavigation[12]Theproposedarchitectureisanextensionofthecontrolarchitecturepresentedin[12]