No difference in arousal or cognitive demands between manual and partially automated driving: A multi-method on-road study
DOI: 10.3389/fnins.2021.577418
archive: archived pipeline: cataloged
Summary
On-road study comparing driver arousal and cognitive demands between manual and partially automated driving across four production vehicles (Cadillac CT6, Nissan Rogue, Tesla Model 3, Volvo XC90) on Utah interstate highways. Seventy-one participants (39 younger, M=28.82 yrs; 32 late-middle-aged, M=52.72 yrs) completed manual and partially automated drives while heart rate, RMSSD (HRV), and Detection Response Task (DRT) reaction time were recorded. Linear mixed-effects models showed no significant main effect of automation on heart rate, RMSSD, or DRT reaction time, and no significant automation-by-age or automation-by-vehicle interactions. Bayes Factor analyses (BF = 0.0002, 0.0004, 0.0002 for HR, RMSSD, and DRT respectively) yielded extreme evidence in favor of the null. Self-report indicated drivers could relax under automation but were not bored, anxious, or more disengaged, and showed appropriate caution on curvy/hilly roads and in heavy traffic. Findings replicate and extend Lohani et al. (2020) pilot, suggesting partial automation as implemented in these production vehicles produces arousal and cognitive demands comparable to manual driving on highway segments.
Key finding
No detectable differences in arousal (heart rate), parasympathetic activity (RMSSD), or cognitive demands (DRT reaction time) between manual and partially automated driving across four production vehicles and two age cohorts; Bayes Factors near 0.0002 provide extreme evidence for the null hypothesis on highway driving.
Methodology
Within-subjects on-road study. 71 licensed drivers (39 younger; 32 late-middle-aged) drove four partially automated production vehicles (Cadillac CT6, Nissan Rogue, Tesla Model 3, Volvo XC90) on Utah interstate highways under both manual and partial-automation (lane centering + adaptive cruise control) modes. Measures: ECG-derived heart rate and RMSSD, and tactile DRT (vibrotactor on left bicep, microswitch on finger, 3-5 s ISI). Analysis: linear mixed-effects models with automation, age, and vehicle as fixed effects and participants as random intercepts, plus Bayes Factor comparisons of full vs restricted (no-automation) models.
Sample size: N=71 (39 younger M=28.82 yrs; 32 late-middle-aged M=52.72 yrs)
Quality score: 5 / 5