Discriminant Model of Driving Distraction During Mobile Phone Conversation Based on Eye Movements

In order to investigate the characteristics of drivers’ eye movements during distracted driving caused by mobile phone conversation and establish a driving distraction discriminant model, a driving simulation experiment was conducted. The eye movement index data were collected by eye tracker under different traffic scenes which include normal driving and perform simple or complex conversation secondary task on the urban road and freeway, then the variance analysis was used to analyze the characteristics. Finally, according to the characteristics of drivers’ eye movements, a driving distraction discriminant model based on fisher discriminant analysis was constructed for different road types. The ANOVA results showed that the effectiveness of road type and conversation task on the cumulative proportion of the driver’s focus on the area of interest in the front road is not statistically significant. However, the average duration of the driver’s attention under urban road scene is significantly higher than that of the freeway, and with the increasing of difficulty of driving task, the average duration of attention increased significantly. In addition, the road type and conversation task significantly influenced the change range of pupil area. The accuracy rate of the discriminant model is 75.2% for the driving distraction on urban roads, and 78.3% for the distraction on freeway.