Examining Take-over Request in Automated Driving: An Exploratory Study Focused on Drivers Asleep at the Wheel

Abstract

A key promise of autonomous vehicles (AVs) is that drivers can engage their primary attention on non-driving related tasks (NDRTs). This behavioral paradigm shift requires that AVs provide intelligibility services appropriate to drivers’ in-situ states and in-car activities, especially NDRTs. Using a high-fidelity AV-simulator based on our recent study, in this paper we examined drivers’ conditions for the most preferred NDRTs. With sleeping scenarios demanding the greatest cognitive and physical load of the driver, this study aims to discern how to enhance vehicle intelligibility in Take-Over Request (TOR) situation across varying feedforward timings ((i) 5 s and (ii) 10 s) and modalities (Auditory, Visual, and Auditory + Visual). The driver’s driving performance and subjective responses were collected. Physiological responses were also measured to obtain a complementary indicator of drivers’ conditions. Messages provided within 10 s were predominantly superior. The visual information contributed positively for a manual drive after TOR, especially before 5 s. However, physiological responses were the opposite due to experiencing a blurry vision after sleeping scenario. This study contributes to designing services in the driver’s sleeping situation in TOR.