Design and implementation algorithm of safe driver assistant system based on EOG

Abstract

In order to ensure driving safety, improve the intelligent level of the vehicle control system and realize“keeping hands on the wheel”, a safe driver assistant system (SDAS) based on EOG was proposed. The proposed sys-tem utilized saccade signals which come from bio-electrodes installed around driver's eyes, to generate some control commands when the driver observes different signs located on the head up display (HUD). Furthermore, independent component analysis (ICA) algorithm was used to extract spatial feature parameters of activity-detected EOG signals, and combined with support vector machine (SVM) method to recognize the type of saccade signals, such as up-rolling, left-rolling and right-rolling. Experiments have been carried out in lab environment, and the average correct ratio on 15 sub-jects is 98.43% and 96.0% corresponding to fatigue condition and non-fatigue condition respectively. Experiential results re-veal that the SDAS based on the multi-class saccade signals recognition algorithm presents an excellent classification per-formance.

Keywords

• EOG;saccade signal;independent component analysis;spatial filtering;support vector machine