Study of cognitive distraction detection based on GMA analysis

Driver cognitive distraction (e.g., voice-based IVISs) has become an important and growing safety concern. Cognitive distraction detection presents an important function for driver distraction mitigation systems. We present a new approach to analyze the difference of correlation between eye movement and steering angle to detect the state of cognitive distraction. Meanwhile, we design a secondary task to simulate cognitive distraction in simulated driving. In the end, experimental analysis shows that the variances and mean value of GMA analysis results between eye movement and steering angel vary between different drivers, but the trend of the difference between cognitive distraction driving and normal driving is consistent. Their difference is enough to detect cognitive distraction.