A Systematic Review of Takeover System Design in Highly Automated Driving

As vehicle automation advances toward highly automated driving, takeover systems must address scenarios where users are entirely disengaged from the driving task. Although highly automated driving remains largely experimental, designing and researching robust takeover systems is essential for ensuring safe deployment, fostering user trust, and promoting system acceptance. Existing research has explored individual aspects of takeover design and interaction, but a comprehensive synthesis remains lacking. This paper presents a systematic review of 61 studies between 2015 and 2024, retrieved from the ACM Digital Library, Web of Science, ScienceDirect, and IEEE Xplore. The review is structured across three core dimensions: 1)takeover system design, including takeover budgets, modalities, and prompt information; 2)takeover scenarios, including driver characteristics, non-driving-related tasks, and traffic environments; and 3)takeover evaluation, including methods, measurement, and experimental apparatus. The analysis highlights inconsistencies in the definitions of automation levels and discusses their implications for comparability and validity in highly-automated driving research. It then identifies concrete research priorities: adaptive multimodal takeover systems, unified evaluation metrics, and ethical frameworks to guide human–system collaboration.