Distracted Driving under Angry: A Study on Driving Performance Analysis and Risk Detection

Abstract

Driver distraction and driver anger have a massive impact on driving risk status. It is significant to study the driving risk state for the intelligent transportation system. However, the mechanism of driving risk under different emotional and distracted states is still unclear. This article investigates the driving risk fluctuations under different driver emotions and distractions. Two distraction tasks (a music task and a two-level cognitive distraction task) were designed through the driving simulator. Twenty-four drivers drove on a simulated urban scene with the data recording. We analyzed the lateral and longitudinal manipulation behaviors and the risk compensation mechanism. K-means clustering was used to analyze the probability distribution of driving risk under different driving distracted and emotional states. Three classification models (deep neural networks(DNN), (k-nearest-neighbor(KNN) and Support Vector Machine(SVM)) were used to identify the driving risk. The results indicate that the driving performance under the influence of anger will worsen further. However, it will also concentrate the driver's attention. As for the two-level cognitive distraction tasks, the high risk probability of high-level cognitive distraction was lower than the low-level. At the same time, the accuracy rate of the driving risk state identification model based on DNN reached 0.908, which could support the development of the driver state management system in the smart cockpit.