Inferring driver's turning direction through detection of error related brain activity

This work presents EEG-based Brain-computer interface (BCI) that uses error related brain activity to improve the prediction of driver's intended turning direction. In experiments while subjects drive in a realistic car simulator, we show a directional cue before reaching intersection, and analyze error related EEG potential to infer if the presented direction coincides with the driver's intention. In this protocol, the directional cue provides an initial estimation of the driving direction (based on EEG, environmental or previous driving habits), and we focus on the recognition of error-potentials it may elicit. Experiments with 7 healthy human subjects yield an average classification 0.69 ± 0.16, which confirms the feasibility of decoding these signals to help estimating driver's turning direction. This study can be further exploited by intelligent cars to tune their driving assistant systems to improve their performance and enhance the driving experience.