Detecting Driver Sleepiness Using Consumer Wearable Devices in Manual and Partial Automated Real-Road Driving
DOI: 10.1109/tits.2021.3127944
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study investigates the viability of using consumer wearable devices to detect driver sleepiness during both manual and partial automated driving on real roads. With the rise of automated driving systems, traditional sleepiness detection methods based on driving performance or facial cues may become ineffective as driver engagement decreases. The authors propose heart rate variability (HRV) as a robust physiological alternative, aiming to determine if partial automation alters the relationship between HRV and sleepiness and to evaluate the classification accuracy of HRV-based models in real-world conditions. The research involved 43 participants who completed four driving trials each: manual and partial automated driving under both alert (daytime) and sleep-deprived (nighttime) conditions. Data were collected over 180 km of real-road driving using commercial Polar H10 chest bands and Garmin sports watches to record heart rate. Subjective sleepiness was measured every five minutes using the Karolinska Sleepiness Scale (KSS), with scores above 7 defined as severe sleepiness. The study extracted time, frequency, and nonlinear HRV features from five-minute epochs preceding each KSS report. An AdaBoost binary classifier was developed to distinguish between alert and sleepy states, employing leave-one-subject-out cross-validation to assess generalizability. Two personalization strategies were tested: baseline correction using initial driving data and personalized calibration using labeled data from one day to predict the other. The results indicate that partial automated driving has a negligible impact on the relationship between HRV and sleepiness. Significant differences in HRV parameters, such as reduced heart rate and increased overall variability, were observed during severe sleepiness episodes regardless of the driving mode. A classifier using only HRV features achieved an area under the curve (AUC) of 0.76. Performance improved to an AUC of 0.88 when HRV features were combined with easily accessible contextual data, specifically driving time and day/night information. Personalization significantly enhanced detection accuracy; baseline correction using a 60-minute window improved results, while personalized calibration yielded the highest performance, though it requires prior labeled data. The study concludes that commercial wearable heart rate monitors are viable tools for assessing driver sleepiness in both manual and partial automated driving scenarios. The findings suggest that HRV-based detection systems do not require separate models for automated driving modes, simplifying implementation. Furthermore, the integration of simple contextual features like time of day and driving duration substantially boosts classification accuracy. This work supports the development of unobtrusive, physiology-based driver monitoring systems that remain effective as vehicles transition toward higher levels of automation, where traditional behavioral indicators are less reliable.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-08-09 |
| archive | success | openalex | — | — | 5 | 2026-08-09 |
| extract | success | cached | — | — | 3 | 2026-08-10 |
| clean | success | clean | — | — | 1 | 2026-08-09 |
| chunk | success | chunk | — | — | 1 | 2026-08-09 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-08-09 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.6-27b-nvidia | summ-v5 | 2 | 2026-08-10 |
| tag | success | vector_similarity | — | — | 10 | 2026-08-11 |
| verify | partial | — | — | — | 1 | 2026-08-10 |
Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified_with_issues.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- drowsiness detection algorithms
- drowsiness
- workload measurement
- truck driver fatigue
- sleep deprivation
- microsleep
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: physiological data
- Methodological Resource: validation psychometrics, tool software