Thispaperproposesanewasymptoticattitudetrackingcontrollerforanunderactuated3-degree-of-freedom(DOF)laboratoryhelicoptersystembyusinganonlinearrobustfeedbackandaneuralnetwork(NN)feedforwardterm.Thenonlinearrobustcontrollawisdevelopedthroughamodifiedinner-outerloopapproach.TheapplicationoftheNN-basedfeedforwardistocompensateforthesystemuncertainties.Theproposedcontroldesignstrategyrequiresverylimitedknowledgeofthesystemdynamicmodel,andachievesgoodrobustnesswithrespecttosystemparametricuncertainties.ALyapunov-basedstabilityanalysisshowsthattheproposedalgorithmscanensureasymptotictrackingofthehelicopter'selevationandtravelmotion,whilekeepingthestabilityoftheclosed-loopsystem.Real-timeexperimentresultsdemonstratethatthecontrollerhasachievedgoodtrackingperformance.