Improving takeover performance during cognitive distraction with an adaptive gaze-based HMI

Automated driving (AD) (i.e., SAE level 3 AD) allows drivers to do something else, yet they can be required to takeover when the AD ends. We tested how an eye-gaze and situation adaptive human-machine interface (HMI) could improve take-over performance when drivers could be cognitively distracted by a conversation. We hypothesized that the eye-gaze and situation adaptive HMI would improve the reconstruction of situation awareness (SA) (the fact of understanding a situation and being able to anticipate the next events) and thereby take-over performance. We conducted a driving simulator experiment with 42 participants (14 per HMI condition: control with a basic HMI, situation adaptive HMI, eye-gaze and situation adaptive HMI) who drove on the highway with an SAE level 3 AD activated and experienced take-over requests. The eye-gaze and situation adaptive HMI improved: (i) SA reconstruction since participants presented higher gaze dispersion, faster RT to the left mirror and spent more time looking at the left mirror, which indicated more visual searches and more information gathered during the take-overs; (ii) take-over performance and safety with increased time to collision and fewer collisions. Our HMI therefore, exhibited promising results. Overall, adaptive HMIs seem to be a good option to ensure better takeover performanceand  safety compared to non-adaptive HMIs.