Visual and cognitive demands of CarPlay, Android Auto, and five native infotainment systems
archive: archived pipeline: cataloged verified
Abstract
Compared workload of using in-vehicle infotainment systems (IVIS) in five OEM vehicles against CarPlay and Android Auto used in the same vehicles. On-road testing with 24 participants per vehicle configuration found workload was significantly greater for the embedded native OEM systems than for CarPlay or Android Auto; CarPlay and Android Auto did not differ in overall demand but showed task-specific strengths and weaknesses.
Summary
On-road study comparing cognitive, visual, subjective, and overall workload of five 2017-2018 OEM native infotainment systems (HondaLink, SYNC 3, MyLink, UVO, Uconnect) against Apple CarPlay and Android Auto running on the same head units. 64 participants drove a 2-mile suburban route at 25 mph while completing audio entertainment, calling/dialing, text messaging, and navigation tasks via center stack or auditory/vocal modes; 24 participants were tested in each vehicle configuration with a planned-missing 5x3x4x2 design. Demand was indexed using Detection Response Task (DRT, ISO 17488) reaction time and hit rate, NASA-TLX, task interaction time, and an overall composite, each scaled relative to a single-task driving baseline (0.0) and high-demand referents (N-back and SuRT, 1.0). Linear mixed-effects analyses with bootstrapped 95% CIs showed native OEM overall demand significantly above the high-demand referent and significantly above CarPlay and Android Auto, which did not differ from each other but both fell below the referent. Mode of interaction effects diverged from prior work: auditory/vocal interactions were lower demand than center stack overall, with CarPlay favoring center stack and Android Auto favoring voice. Task-by-system interactions: CarPlay was lowest for text messaging; Android Auto was lowest for navigation entry; calling/dialing favored CarPlay over Android Auto. Between-vehicle variability was tighter for Android Auto than for CarPlay or native systems despite identical hardware.
Key finding
CarPlay and Android Auto produced significantly lower overall workload than the embedded portion of native OEM infotainment systems across five 2017-2018 vehicles, while not differing from each other; the two phone-projection platforms have task- and mode-specific strengths (CarPlay better for texting and center-stack interactions; Android Auto better for navigation and voice interactions).
Methodology
On-road study, 5 (Vehicle) x 3 (System: native OEM, CarPlay, Android Auto) x 4 (Task: audio, calling, texting, navigation) x 2 (Mode: auditory/vocal vs center stack) factorial with planned missing data; 64 participants total, 24 per vehicle cell, 2-mile suburban 25-mph route. Workload indexed by DRT (ISO 17488) RT and hit rate, NASA-TLX, task interaction time, and a composite overall demand score, each baseline-corrected and scaled to single-task driving (0.0) and N-back / SuRT high-demand referents (1.0). Linear mixed-effects models with likelihood-ratio tests and 10,000-sample bootstrapped 95% CIs.
Sample size: N=64 (32 female, ages 21-36, M=25); 24 participants tested per vehicle configuration
Quality score: 5 / 5