Effect of cognitive load on drivers' State and task performance during automated driving: Introducing a novel method for determining stabilisation time following take-over of control

Vadim Melnicuk,Simon Thompson,Paul Jennings,Stewart Birrell

DOI: https://doi.org/10.1016/j.aap.2020.105967

Abstract:This research paper explores the impact of cognitive load on drivers' physiological state and driving performance during an automated driving to manual control transition scenario, using a driving simulator. Whilst driving in the automated mode, cognitive load was manipulated using the "N-Back" task, which participants engaged with via a visual display. Results suggest that non-optimal levels of workload during the automated driving conditions impair driving performance, especially lateral control of the vehicle, and the magnitude of this impairment varied with increasing cognitive load. In addition to these findings, the present paper introduces a novel method for determining stabilisation times of both driver state and driving performance indicators following a transition of vehicle control. Using this method we demonstrate that mean and standard deviation of lane position impairments were found to take longer to stabilise following transition to manual driving following a higher level of cognitive load during the automated driving period, taking up to 22 s for driving performance to normalise after take-over. In addition, heart rate parameters take between 20 and 30 s to stabilise following a planned take-over request. Finally, this paper demonstrates how the magnitude of cognitive load can be estimated in context of automated driving using physiological measures, captured by consumer electronic devices. We discuss the impact our findings have on the design of SAE Level 3 systems. Relevant suggestions are provided to the research community and automakers working on future implementation of vehicles capable of conditional automation.