Discrimination Between Normal Driving and Braking Intention from Driver’s Brain Signals

Abstract

Advanced driver-assistance systems (ADAS) are in-car technologies that record and process vehicle and road information to take actions to reduce the risk of collision. These technologies however do not use information obtained directly from the driver such as the brain activity. This work proposes the recognition of brake intention using driver’s electroencephalographic (EEG) signals recorded in real driving situations. Five volunteers participated in an experiment that consisted on driving a car and braking in response to a visual stimulus. Driver’s EEG signals were collected and employed to assess two classification scenarios, pre-stimulus vs pos-stimulus and no-braking vs brake-intention. Classification results showed across-all-participants accuracies of 85.2 ± 5.7% and 79 ± 9.1%, respectively, which are above the chance level. Further analysis on the second scenario showed that true positive rate (77.1%) and true negative rate (79.3%) were very similar, which indicates no bias in the classification between no-braking vs brake-intention. These results show that driver’s EEG signals can be used to detect brake intention, which could be useful to take actions to avoid potential collisions.