Steering Entropy Changes as a Function of Microsleeps

Paul, Amit; Boyle, Johnell O; Boer, Erwin R; Tippin, Jon; Rizzo, Matthew · 2005 · Crossref

DOI: 10.17077/drivingassessment.1196

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 utility of steering entropy (SE) as a metric for detecting decrements in driving performance caused by microsleeps in drivers with obstructive sleep apnea/hypopnea syndrome (OSAHS). Motivated by the significant economic and safety costs associated with sleep-related motor vehicle crashes, the research aims to establish an objective criterion for identifying drowsy drivers. The authors hypothesize that SE, which quantifies the randomness and erratic nature of steering inputs, can effectively capture the aggressive corrective actions drivers make after losing attention during microsleeps, particularly on curved road segments where steering demands are higher. The methodology involved 24 legally licensed adult drivers (aged 30–65) with medically diagnosed OSAHS. Participants completed three laps of a simulated two-lane highway using the SIREN driving simulator, which included straight segments and curves with 600-meter radii. Microsleeps were identified via electroencephalography (EEG) criteria, defined as episodes lasting 3–14 seconds characterized by specific changes in alpha and theta brain wave activity. The study employed a case-crossover design, comparing steering entropy during microsleep episodes against matched control periods immediately preceding and following each event. SE was calculated from steering angle data, measuring the deviation between actual and predicted steering angles. Statistical analysis utilized a repeated measures mixed-effects ANOVA to assess differences across microsleep periods (pre, during, post) and road types (straight, curve). The results indicated that steering entropy significantly differed across the three microsleep periods, with SE values higher during microsleep episodes (mean SE = 0.48) compared to pre-microsleep periods (mean SE = 0.40). Post-microsleep SE (mean SE = 0.45) remained elevated relative to the pre-period. Road segment type also significantly affected SE, with higher values observed on curves compared to straight roads. Notably, drivers experiencing microsleeps on curved segments exceeded established safety thresholds for steering entropy both during and after the microsleep events. The data revealed that SE did not decrease during microsleeps, contrary to expectations for inaction, but rather increased, reflecting the transitional state of consciousness and erratic motor behavior associated with OSAHS. The findings suggest that steering entropy is a sensitive metric for detecting erratic steering behavior and increased workload associated with microsleeps. The study concludes that SE can serve as a real-time marker for abnormal driver physiology, potentially enabling the development of alerting systems for drowsy drivers. By capturing the aggressive corrective steering actions that follow lapses in attention, SE offers a promising tool for improving driver assessment and safety interventions, particularly in high-demand driving scenarios such as navigating curves. Future research is recommended to correlate SE with conventional performance measures like lane deviation to further validate its application in safety countermeasures.

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.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• microsleep
• drowsiness
• drowsiness detection algorithms
• steering pattern
• sleep deprivation
• vigilance

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, behavioral performance data
• Theoretical Contribution: computational model