Combined regression and classification approach for prediction of driver´s braking intention

Recent studies for driving assistant system have been concerned with driver´s convenience and safety. Especially, neurophysiological studies were employed to develop the novel driving assistant technologies for driver´s safety. These studies verified that neurophysiological characteristics could be used for detection of emergency situations during simulated driving. However, it is impossible to control the vehicle spontaneously using previous approach. In this article, the method for decoding of driver´s braking intention spontaneously is proposed to predict the amount of braking continuously based on analysis of neural correlates. The prediction results based on Kernel Ridge Regression (KRR), linear regression, and combined linear regression and classification approaches are compared and evaluated by the normalized root-mean square error (NRMSE) and one-way ANOVA for statistical test.