一类非线性时间缩放函数用于沿指定路径进行平滑优控制课件.pdfVIP

2015IEEE/RSJInternationalConferenceonIntelligentRobotsandSystems(IROS)

CongressCenterHamburg

Sept28-Oct2,2015.Hamburg,Germany

AClassofNon-LinearTimeScalingFunctionsForSmoothTime

OptimalControlAlongSpecifiedPaths

ArunKumarSinghandK.MadhavaKrishna

—ComputingtimeoptimalmotionsalongspecifiedAsshownin[5],[6],suchrepresentationleadstoconvexity

pathsformsanintegralpartofthesolutionmethodologyforandsparsityinthestructureoftheresultingoptimization.

manymotionplanningproblems.Conventionally,thisopti-Convexoptimizationsarenotonlycomputationallyfast,but

malcontrolproblemissolvedconsideringpiece-wiseconstant

parametrizationforthecontrolinputwhichleadstoconvexityalsovarioushighfidelityopensourcesolverslike[7]existfor

andsparsityintheoptimizationstructure.However,italsorapidprototyping.Butthiscomputationalgainprovidedby

resultsindiscontinuouscontroltrajectorywhichisdifficulttopiece-wiseconstantformcomesatthecostofdiscontinuous

track.Thus,inthispaperwerevisitthistimeoptimalcontrolcontroltrajectorywhicharedifficulttotrack[8].Although

problemwiththeprimarymotivationofensuringahighdegree[9]usespiece-wiseconstantcontrol,itincludesadditional

ofsmoothnessintheresultingmotionprofile.Inparticular,we

solveitwithcontinuityconstraintsincontrolandhigherorderjerkconstraintstosmoothenthetransitionofcontrolinputs

motionderivativeslikejerk,snapetc.Itisclearthatsuchbetweentheumandminimumbound.

constraintswouldnecessitatetheuseoftimevaryingcontrolInthispaper,