Design and implementation of virtual driving system fusing driver´s cognitive and operating characteristics

Mental fatigue affects driver´s alertness and ability to drive safely, which is easy to cause traffic safety problems. The aim of this paper is to establish a driving simulation system based on virtual-reality technology to offer an environment for continuously monitoring driver´s drowsiness level. By using 3DS max software for the modeling of car parts, driving scenes and Virtools for the establishment of automobile´s motion system, this system can provide a driving simulation experiment platform so as to collect EEG data and other parameters during the driving process. Then, through studying different classification methods of different fatigue status, the SVM algorithm is chosen for the assessment of mental fatigue. So we could explore the relationship between the collected EEG data and driving operation data. At last, we use the EEG feature vectors derived from CSP algorithm as input data and the level of fatigue obtained by using the driving operation data as the output to train the fatigue detection system and test it with corresponding data. The result shows that combined CSP with Wavelet-packet transform algorithm, the fatigue status classification accuracy is 94.2529% (82/87).