Analyzing Driver Emotion & Behavior during Peak and Off-Peak Conditions

Abstract: The increasing urbanization and growth in settlement sizes are driving the transportation demand in India’s cities. As a result, vehicles population is increasing, giving rise to greater congestion, and bringing along road safety challenges. Global road accident fatalities were estimated to be 11.1 lakhs in 2021, with a majority of them recorded in middle and low-income countries. India witnessed a total of 4,61,312 road accidents in the year 2022, with an annual growth rate of 11.9%. Despite several enforcement measures taken by the government and law enforcement agencies, the road accident severity has been consistently increasing over the past few years. Research indicates that driver distraction and human errors are the significant causal factors of global road accidents. Globally, research on driving behavior started receiving prominence in the early 2000s, where initially “Driver Behavior Questionnaires” were used to understand the driving behavior. Since then, advancements in technology, i.e., driving simulators, have gained prominence in research. Yet, they come with certain disadvantages, like inability to accurately represent the real-world scenarios. Naturalistic driving behavioral studies have evolved as an effective way of understanding driving behavior with real time data capture. The predominant research on naturalistic driving behavior is being conducted in the United States. USA has implemented distracted driving laws targeting the novice drivers, which have resulted in significant reduction in fatalities, especially among young drivers.
This study employs naturalistic driving data of 100 minutes of 4 drivers, using Dashcam (Emotion and Distraction), VBox (Vehicle Kinematics) and Smart watch (Heart Rate). As the vehicle kinematics are influenced by many factors in real world, heart rate data collected using smart watches is used to validate the influence of driver emotion. The data set captured temporal data of drivers’ emotions during vacation and peak hours in an academic residential campus setting. The data was analyzed using OpenCV and Deepface facial recognition algorithms. Central tendency, Analysis of Co-variance (ANOVA), Correlation and Trend analysis were used to estimate the impact of driver emotion on vehicle kinematics. Study results show that drivers predominantly depict neutral emotion in low traffic conditions, but tends to become angry in congested situations. This could be a result of increased near-conflicts and resultant frequent braking behavior. The average speed attained under low traffic conditions is 24.6 km/h, which decreased to 19.4 km/h during peak hour. Likewise, average deceleration also decreased from -0.76g to -0.48g during a congested drive, indicating reduced braking intensity in such conditions. Finally, the results show that driver emotion has a significant impact on the acceleration and braking of the vehicle. The findings suggest that neutral driver emotion leads to a stable driving behavior, this could assist drivers in adopting self-regulatory mechanisms that overcome cognitive decision making under the influence of emotions.