Abstract
The advanced airbag system, which provides superior airbag inflation control in the event of collision, requires an occupant classification sensor. This sensor classifies the passenger seat occupant into either adult or child including a child seat, and suppresses the airbag inflation accordingly. An occupant classification sensor, embedded in the seat cushion, measures occupant pressure distribution via an array of pressure-sensitive cells and detects children and child seats by processing data using pattern recognition. This paper proposes applying Support Vector Machine (SVM), a completely new statistical pattern-recognition method, to the algorithm of the occupant classification sensor. Our experimental results confirmed that the SVM algorithm can significantly reduce error rate, compared to conventional methods.