Developing Standard Metrics for Measuring Takeover Performance in Highly Automated Driving (SAE Level 3 and above)
Xi Jessie Yang
Assistant Professor of Industrial and Operations Engineering, College of Engineering and Assistant Professor of Information, School of Information
Professor of Mechanical Engineering and Professor of Electrical Engineering and Computer Science, College of Engineering
Associate Professor of Information, School of Information
Associate Research Scientist, Behavioral Sciences, University of Michigan Transportation Research Institute and Adjunct Associate Research Scientist, Psychology, College of Literature, Science, and the Arts
A particular concern with Level 3 automation is the takeover transition from the automated vehicle (AV) to the driver. As the driver is not always required to monitor the environment, they may become increasingly decoupled from the driving task and have difficulty in taking over control at a moment’s notice.
Takeover performance consists of two aspects: timeliness and quality. Timeliness is commonly measured using takeover time. However, for measuring takeover quality, studies use different metrics at their own discretion. The lack of a standard makes it extremely difficult to consolidate findings from prior studies and to compare the effectiveness of different designs. We believe it is crucial to develop a standard set of metrics for measuring takeover quality.
The proposed effort is intended to address current gaps with two specific aims:
(1) To develop a standard set of metrics for measuring takeover quality (and also, takeover performance). This standard will be proposed, tested and refined via a large-scale literature survey and a series of human-in-the-loop experiments.
(2) To quantify the relationship between takeover readiness, timeliness and quality. The research team has an ongoing Mcity funded project that examines drivers’ takeover readiness. The proposed effort will allow us to further examine how takeover readiness affects takeover timeliness and quality.