Answers are grounded in this paper and cite their sources.
Background: Human factors are critical for road safety. Driving simulators offer controlled, repeatable conditions for studying driver performance, but it is unclear which simulator findings are supported by real-world evidence. Objective: This systematic review examines (i) how far behavior in driving simulators is supported by real-world or naturalistic evidence, and (ii) how driver responses to hazards and takeover requests are measured, and how often they, and the driver-state and physiological indicators linked to them, have been examined in real-world conditions. Methods: Google Scholar and ScienceDirect were searched (last search 19 October 2026) for English-language publications from 2014 to 2026, complemented by citation searching; PRISMA 2020 was used as a reporting guide. Seventy-eight publications were included and synthesized thematically. No formal risk-of-bias tool was used. The review was not registered and had no external funding. Results: Simulated and real driving agreed more for speed, lateral position, and following distance than for workload or absolute response magnitudes. Faster takeover was not always safer, so response speed and control quality should be reported separately. Physiological indicators are associated with cognitive load but vary between drivers. Comparisons of simulated and real driving remain few. Conclusions: Matched real and simulated scenarios, explicit response definitions, and synchronized behavioral and physiological measures are needed to make simulator findings more reliable and support driver monitoring, advanced driver assistance systems (ADAS), and road safety.
No takes yet. Share an insight, caveat, or question.
Bobeș et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: