Driver Recognition System Using FNN and Statistical Methods

Abstract

Advancements in biometrics-based authentication have led to its increasing prominence and are being incorporated into everyday tasks. Existing vehicle security systems rely currently on electronic alarm or smart card systems. A biometrie driver recognition system utilizing driving behavior signals can be incorporated into existing vehicle security system to form a multimodal identification system and offer a higher degree of protection. The system can be subsequently integrated into intelligent vehicle systems where it can be used for detection of any abnormal driver behavior with the purposes of improved safety or comfort level. In this chapter, we present features extracted using Gaussian Mixture Models (GMM) from accelerator and brake pedal pressure signals, which are then employed as input to the driver recognition module. A novel Evolving Fuzzy Neural Network (EFuNN) was used to illustrate the validity of the proposed system. Results obtained from the experiments are compared with those of statistical methods. They show potential of the proposed recognition system to be used in real-time scenarios. A high identification rate and the low verification error rate were indicated considerable difference in the way different drivers apply pressure to the pedals.