Toward Estimation of Abnormal Brake in Autonomous Vehicles from Electroencephalogram and Heart Rate Interval

The development of automated driving technology has concentrated on safety aspects such as preventing traffic accidents and not focusing on ride comfort. To realize a reliable technology, it is necessary to consider the ride comfort since automated vehicles use computer control for all operations such as steering wheel and gas pedal. In this paper, we hypothesize that abnormalities in braking timing were reflected in the physiologi-cal signals. The abnormal braking timing can be identified from the characteristics of EEG and heartbeat interval. We analyzed the EEG and heart rate intervals of normal and abnormal brake timing created by the simulator to verify the hypothesis. We classified the brake condition by a support vector machine. As a result of using the features of the EEG and heartbeat intervals to discriminate the abnormal brake using a support vector machine, four out of the nine experimental participants achieved an average correct answer rate of more than 80 %. Therefore, it was found that the abnormal brake could be estimated from EEG and heartbeat interval features under the presentation of driving videos with different braking timing.