Impact of Buffer Time on Driving Behavior in Level-3 Autonomous Vehicles

Abstract This study examines how buffer time impacts the behaviors of 20 young drivers in terms of takeover duration and quality, visual attention and cognitive effort, and emotional states preceding, during, and subsequent to take over requests (TORs) upon system failure in level 3 autonomous vehicles occurring just before intersections. Two groups of buffer times of 8 and 4 s were compared to investigate the impacts of buffer times on driving behavior, eye gaze patterns, and EEG data related to emotional states. Three separate MANOVA tests were conducted to analyze how buffer time impacted these constructs. The individual MANOVA was used for each construct, which included various dependent variables, and included two independent variables (8 s and 4 s groups, referred to as 8SG and 4SG respectively) and multiple dependent variables. The results showed that buffer times do not affect drivers’ emotional states and the duration and quality of takeovers. On the other hand, this approach has a statistically significant effect on drivers’ attention given to traffic signals. Drivers’ emotional states (excitement, engagement, stress, relaxation, interest, and focus) are not influenced by buffer time upon system failure. Additionally, when drivers are provided with earlier warnings, they pay further attention to the traffic signal. Finally, drivers take control of their vehicles faster with an earlier buffer time than with a later one.