Tracking Control ofMobile Robots BasedonImproved RBFNeural Networks

A control schemefordynamic tracking ofmobile robots ispresented, whichintegrates avelocity controller based onbackstepping techniques anda torquecontroller basedon improved RBF neuralnetworks. Becausethetorquecontrol strategy derived fromsliding modesdepends onthedynamics of mobile robots, therobustness ofthesystem cannot beguaranteed duetotheuncertainties ofrobotdynamics. Inordertodecrease theimpact oftheuncertainties andimprove therobustness ofthe system, improved RBF neural networks aredesigned online to modelthedynamics ofmobile robot. Thusthetorque controller basedonsliding modeiscomposed ofaneural network controller andarobust compensator. Simulations demonstrate theefficacy oftheproposed system forrobust tracking ofmobile robots. IndexTerms- Mobilerobots, dynamic tracking, improved RBFneural networks, sliding mode.