AnAdaptable ObjectClassification Framework

The classification of observedobjects in the vehicle's environment isnecessary forseveral active safety systems. A frameworkfortheobject classification taskis introduced. Theclassification benefits frompattern classification aswellasfromrulebasedapriori knowledge. Theframework can servedifferent applications atthesame time.A new approach isapplied toadapttheoutput foreachapplication toits special requirements. Thisadaptation consumesonlyverylittle processing timeandcanbeperformed formultiple applications without affecting theframework's realtimeproperties. The practical usageof theframeworkisillustrated by the classification ofmeasurementsofa Laserscanner, butthe framework isalsoapplicable forother typesofsensors.