Examining the effect of infotainment auditory-vocal systems' design components on workload and usability
DOI: 10.1016/j.trf.2019.02.006
archive: archived pipeline: cataloged verified
Summary
On-road study of twelve 2016-2017 production vehicles examining how auditory-vocal infotainment design components (delay time, menu depth, system accuracy) predict driver workload and usability. 120 drivers (24 per vehicle) completed voice-dialing and voice-tuning tasks on a residential route while NASA-TLX, SUS, task duration, and verbal sentiment were collected. For voice dialing, total delay time predicted task duration (R^2=.39) and combined with delay-time variability predicted workload (R^2=.57). For voice tuning, total delay time and menu depth predicted workload (R^2=.52), complexity (R^2=.77), ease-of-use (R^2=.73), SUS (R^2=.81), and sentiment polarity (R^2=.73). System accuracy was not a significant predictor. A regression model from the original 12 vehicles was validated against seven additional vehicles with no significant difference between observed and predicted task durations.
Key finding
Voice-system delay time and menu depth, but not recognition accuracy, drive increased mental workload and longer task durations, with delay time alone explaining a substantial share of variance in usability and sentiment ratings across twelve production infotainment systems.
Methodology
On-road study using 12 production vehicles (later validated against 7 more). Each system characterized by total delay time, delay-time SD, menu depth, and voice-recognition accuracy. 120 participants drove a 25 mph residential route while issuing voice-dialing and voice-tuning commands. Hierarchical regression predicted task duration, NASA-TLX workload, SUS usability, and sentiment polarity from system design metrics.
Sample size: 120 drivers (54 female), age 21-36 (M=25); 24 per vehicle across 12 vehicles; planned-missing design; 7 additional vehicles for validation
Quality score: 5 / 5