From the eeg signals obtained from the drivers with the drive simulator, the emergency braking situation is estimated by artificial neural networks

Increasing numbers of automobiles in the world are accompanied by an increase in security measures. Previous studies on the control of emergency braking situations using EEG signals have used Sequential Forward Floating Search (SFFS) method and Event Related Potentials (ERP) signs of EEG activities. With this study, it is possible to estimate the emergency braking situations by using Electroencephalogram (EEG) signals from the drivers during driving with the driver simulation system. In this study, the features of the EEG signals received from the drivers during driving with the drive simulator system were obtained by using the pwelch method, and the signals are divided into delta, tetra, and beta lower frequency bands. The signals belonging to the sub frequency bands were trained using Artificial Neural Networks (ANN). During the training of EEG signals, Feed Forward Backpropagation and Learning Vector Quantization (LVQ) methods are used at the ANN. In the sub frequency bands of the EEG signals from the drivers, the prediction of the emergency brake conditions of the leading vehicle has been realized. At the same time, predictions were made for the increase and decrease in brake deviation data of the leading vehicle. The network was trained with the Mean Squared Error (MSE) 0,005 values. Estimates of the data were made with 70% accuracy in the delta sub frequency band and test data applied to the network. However, when the drivers pressed the gas pedal and did not push it, the EEG data was estimated with YSA. MSE value of 0,000001 was obtained in network training and 90% of the test data were correctly estimated.