BROWSE
Detail
Conference Paper
Clustering Attentional Types in Conditional Automated Driving for Adaptive Take-Over Request Design
- Authors
- Mi Chang, Eun Hye Jang, Woojin Kim, Jiwoo Han, Daesub Yoon, Yang Koo Lee, Kezia Amanda Kurniadi
- Issue Date
- 2026-07
- Citation
- ACM SIGGRAPH 2026, pp.1-3
- Publisher
- ACM
- Language
- English
- Type
- Conference Paper
- Abstract
- In SAE Level 3 conditional automated driving, a driver’s attentional state before a Take-Over Request (TOR) critically shapes take-over quality, yet Driver Monitoring Systems rely on forward-gaze duration even though sustained forward gaze can coexist with mind-wandering. We applied K-Means clustering to 18 gaze and head-movement features from 93 drivers in a CARLA simulator and identified three attentional types: Focused, Exploratory, and Disengaged. The Focused type showed the highest forward-gaze ratio yet significantly lower situational awareness than the Exploratory type under high-demand conditions, a Forward Gaze Paradox that gaze-duration metrics cannot capture. The Disengaged type maintained near-ceiling physical readiness despite the lowest situational awareness, a cognitive-behavioral dissociation invisible to gaze-based monitoring. These types ground adaptive TOR interface design.
- KSP Keywords
- Automated driving, Driver monitoring, Gaze-based, High-demand, Interface design, Monitoring system, Situational Awareness, k-means clustering, level 3, movement features, take-over request
This work is distributed under the term of Creative Commons License (CCL)
(CC BY)
(CC BY)
