Towards Safe and Comfortable Vehicle Control Transitions: A Systematic Review of Takeover Time, Time Budget, and Takeover Performance
URL: http://arxiv.org/abs/2507.22262v1
archive: archived pipeline: in_review verified
Abstract
Conditionally automated driving systems require human drivers to disengage from non-driving-related activities and resume vehicle control within limited time budgets when encountering scenarios beyond system capabilities. Ensuring safe and comfortable transitions is critical for reducing driving risks and improving user experience. However, takeovers involve complex human-vehicle interactions, resulting in substantial variability in drivers' responses, especially in takeover time, defined as the duration needed to regain control. This variability presents challenges in setting sufficient time budgets that are neither too short (risking safety and comfort) nor too long (reducing driver alertness and transition efficiency). Although previous research has examined the role of time budgets in influencing takeover time and performance, few studies have systematically addressed how to determine sufficient time budgets that adapt to diverse scenarios and driver needs. This review supports such efforts by examining the entire takeover sequence, including takeover time, time budget, and takeover performance. Specifically, we (i) synthesize causal factors influencing takeover time and propose a taxonomy of its determinants using the task-capability interface model; (ii) review existing work on fixed and adaptive time budgets, introducing the concept of the takeover buffer to describe the gap between takeover time and allocated time budget; (iii) present a second taxonomy to support standardized and context-sensitive measurement of takeover performance; (iv) propose a conceptual model describing the relationships among takeover time, time budget, and performance; and (v) outline a research agenda with six directions.
Summary
Systematic review of 100 articles (2010-2025, Scopus+Web of Science, PRISMA filtering of 277 records) synthesizing the takeover sequence in conditionally automated driving across three constructs: takeover time, time budget, and takeover performance. The authors propose two taxonomies (one for takeover-time determinants based on the task-capability interface model, one for takeover-performance indicators spanning vehicle-related driving-quality measures and human-related psychophysiological/behavioral measures) and introduce 'takeover buffer' (time budget minus takeover time) as the appropriate measure linking budget to performance. A qualitative hypothesis is formulated about how takeover time, time budget, and takeover performance interrelate, and a six-gap research agenda for adaptive time-budget design is outlined.
Key finding
Sufficient time budgets must be balanced - too short compromises safety/comfort, too long reduces alertness - and require an adaptive design that integrates both vehicle-side and human-side takeover-performance indicators rather than the currently imbalanced vehicle-dominant measures.
Methodology
systematic_review
Sample size: 100 articles synthesized after PRISMA filtering of 277 initial records (185 Scopus + 92 Web of Science, 2010-2025)