We extend this concept to robotic hands and show how a similar dimensionality reduction can be de64257ned for a number of different hand models This framework can be used to derive planning algorithms that produce stable grasps even for highly compl ID: 76772
Download Pdf The PPT/PDF document "Dimensionality reduction for handindepen..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Dimensionalityreductionforhand-independentdexterousroboticgraspingMateiCiocarlieCoreyGoldfederPeterAllenInthispaper,webuilduponrecentadvancesinneuroscienceresearchwhichhaveshownthatcontrolofthehumanhandduringgraspingisdominatedbymovementinacongurationspaceofhighlyreduceddimensionality.Weextendthisconcepttorobotichandsandshowhowasimilardimensionalityreductioncanbedenedforanumberofdifferenthandmodels.Thisframeworkcanbeusedtoderiveplanningalgorithmsthatproducestablegraspsevenforhighlycomplexhanddesigns.Furthermore,itoffersaunied Gripper4 DOF Barrett4 DOF DLR12 DOF Robonaut14 DOF Human20 DOF Fig.3.Eigengraspplannertestusing5handmodelstograspeachof6objectsoftheresults,gure3presentsthebesthandposturefoundbytheplannerwithoutanyadditionalrenements.Wenotethat,inmostcases,planninginthereducedspacespannedbyonlytwoeigengraspsdoesnotresultinaposturewheretherobotichandconformsperfectlytothesurfaceoftheobject.However,theresultisoftencloseenoughtosuchaposturethatastablegraspcanbeobtainedbyusingsimpleheuristics.Onepossibleheuristicinvolvesclosingeachngeruntilcontactwitheithertheobjectoranotherngerpreventsfurthermotion.Thismethodproducesaforce-closuregraspin23outofthe30casespresentedingure3.Fortheresultspresentedingure3,wehavespecieddesiredcontactlocationsontheentiresurfaceoftheroboticpalm.However,itisalsopossibletouseonlyasubsetofthese.Forexample,wecanchoosetouseonlythengertipcontacts,thusremovingtherequirementofwrappingthehandaroundtheobject.Inthiscase,thegraspqualitycomponentofourenergyfunction(sectionIII-B)takesvitalimportance,asitisgenerallyeasytosimplyplacethengertipsontheobjectsurfacewithoutnecessarilycreatingastronggrasp.Thisapproachcanberegardedasanattempttondgoodmanipulationgraspsand,especiallyinthecaseofrobotichandsequippedwithhuman-likengertips,itproducesstableresults.Examplesareshowningure4.a.fortheDLR,RobonautandHumanhandmodels,withthenotethatallpresentedgraspshaveforce-closure.Finally,wehaveusedthemethodpresentedinthispapertoplangraspsinthepresenceofobstacles.Figure4.b.showsasituationinwhichatablesurfacepreventstheexecutionofthebestgrasps,thusforcingthealgorithmtondalternativesolutions.Theonlyadditionalcostincurredbythegraspplanneristhatofcollisiondetectionagainsttheobstacleforeachnewlygeneratedstate.Again,allthegraspspresentedintheimagehaveforce-closure.Oneobjectinourtestsetthatrequiresadditionalconsider-ationisthetoyairplanemodel.Itdiffersfromtherestofthesetinthesensethatitcannotbewellapproximatedusingasingleconvexcomponent,asitisthesumofanumberofdominantshapes(fuselage,wings,etc.).Thegraspplannerthatwehavepresentedposesnohardconvexityconstraintontheobject,anditproducesanumberofconvincinggraspsontheairplanemodel.However,anintuitivemethodtosimplify