Andynamicsystemforreal-timeobstacleavoidancepathplanningofredundantrobotsisconstructedinthispaper.Firstly,theinter-framedifferencemethodisusedtoidentifythemovingtargetandtocalculatethetargetarea,thenonthebasisofcolorfeaturesandgradientfeaturesextractedfromthetargetarea,thefeaturefusionCam-Shiftmeanshiftalgorithmisusedtotracktarget,improvingtherobustnessofthetrackingalgorithm.Secondly,aparalleltwo-channeltargetidentificationandlocationmethodbasedonbinocularvisionisproposed,updatingthetarget’sthree-dimensionalinformationinrealtime.Then,adynamiccollision-freepathplanningmethodisimplemented:thesafetyrodsareremovedthroughtheintersectiontest,andtheminimumdistanceisderiveddirectlybyusingthecoordinatevaluesofthetargetinthelocalcoordinatesystemoftherod.Onthisbasis,theobstacleavoidancegainandescapevelocityrelatedtotheminimumdistanceisestablished,andobstacleavoidancepathplanningisimplementedbyusingthezerospacemappingmatrixofredundantrobot.Experimentsareperformedtostudytheefficiencyoftheproposedsystem.
