Detecting Physiological Changes In Response To Sudden Events In Driving: A Nonlinear Dynamics Approach

In this study, we propose a novel analytic framework to detect emergency braking intentions in driving tasks by capturing the delay in human responses using multi-modal biosensor data, e.g., electroencephalography (EEG) and electromyography (EMG). To quantify the response delay, we consider EEG and EMG signals as a coupled dynamic system and employ a recurrence plot (RP) based approach to characterize the nonlinear dynamics. We then apply the maximally stable extremal regions (MSER) method in computer vision for detecting transition states associated with sudden events (e.g., braking intentions in driving). Our proposed framework is tested on a publicly available dataset of driving experiments. The results demonstrate the effectiveness of our proposed approach for assessing the response delay to reflect the motor control command, which shows that the average response delays to the braking intentions are 300 milliseconds in EEG and 194 milliseconds in EMG prior to the actual emergency braking. The proposed quantification can be employed in driving assistant system for reducing or diminishing potential accidents.