Putting the Brakes on Autonomous Vehicle Control

Abstract

The assimilation of automation in commuter vehicles is rapidly increasing, as too are the concerns with these technologies. Human interaction with autonomous vehicles must be thoroughly researched to understand the quantification and qualification of interactive behaviors with these systems. We developed a study using a high-fidelity driving simulator to mimic probable breakdowns with these systems to better understand the subsequent human responses and to explore the necessary technological requirements to overcome potential problems. 30 participants engaged in a driving scenario switching between manual and autonomous vehicle control. We accounted for individual differences in braking reaction time while simultaneously engaging in a secondary cognitive task during times of autonomous vehicle control. Results show the average RT for baseline scenarios without the cognitive task was 832.1 milliseconds while the average RT for baseline scenarios with the cognitive task was 908.4 milliseconds; a 9.17% significant increase. The average RT for the autonomous scenario was 1357.0 milliseconds; a significant increase of 49.38% over the baseline scenario with the cognitive task that can be attributed to the addition of automation. We found a positive linear correlation of time spent in autonomous control and subsequent braking reaction time. Additionally, cognitive task difficulty, attention allocation, self-reported mental demand, fatigue, and heart rate affect reaction time when cued to take control of the vehicle.