Auditory cues in automated driving: enhancing takeover performance with spatialized looming warnings

In Level 3 automated driving, driver takeover poses a significant safety challenge, especially when drivers are engaged in non-driving-related tasks. Although takeover requests (TORs) often use auditory signals, their ability to convey both urgency and directional hazard information remains underdeveloped. The present study examined whether two features of auditory TORs improve driver takeover performance: a gradually increasing intensity profile (looming) and differences in sound level between the left and right channels (spatialization). These features were designed to convey temporal urgency and hazard direction, respectively. Thirty-five licensed drivers completed a simulated driving task while performing a visually demanding typing task. In a within-subject design, they encountered a rear-end conflict with a lead vehicle and two lateral conflicts involving a crossing pedestrian or a fast vehicle approaching from behind. Intensity profile (looming vs. constant) was manipulated across all conflicts, with spatialization (spatialized vs. non-spatialized) manipulated in the lateral hazards. Results showed that spatialized looming TORs significantly improved takeover performance, consistently promoting earlier control initiation and more efficient takeovers. Specifically, the looming profile provided robust benefits, with spatialization offering additional benefits in several contexts. Subjective evaluations further revealed both benefits and drawbacks: spatialized looming warnings were rated as more effective, informative, and helpful, whereas looming cues were also perceived as more annoying. Overall, this study demonstrates that auditory TORs can effectively convey temporal urgency and hazard direction to enhance takeover performance. Future research should evaluate these warning strategies in more ecologically valid driving environments while optimizing the balance between effectiveness and driver acceptance.