Driver identification and driver's emotion verification using KDE and MLP neural networks

Abstract

A Driver's behavior is a major factor that contributes to the high accidents rates. However, if we are able to identify their behavior, it may be possible for us to detect driving idiosyncrasies that may prevent accidents. Therefore, this paper presents some simple and effective methods for an in-car data acquisition in collecting real time driving data. The data has been classified into three different driver's condition which leads into accident. They are happy expression, talking on the phone and normal driving. These data will be used to investigate the effectiveness of a driver's behavior which focusing on the driver's response towards the brake and gas pedals as well as its rate of change. From these data we will demonstrate simple yet effective technique in driver identification and driver verification. We use the kernel density estimation (KDE) as tools to extract features. Then, we use these features to recognize the emotion of the driver by using multi layer perceptron (MLP) as classifiers. The enhancement of driver's security, safety and comfort driving can be derived trough the performances of the driver's emotion verification which contribute to the development in the area of intelligent vehicle driver verification system.