Multimodal Features for Detection of Driver Stress and Fatigue: Review

Němcová, Andrea; Svozilová, Veronika; Bucsuházy, Kateřina; Smíšek, Radovan; Mézl, Martin; Hesko, Branislav; Belák, Michal; Bilík, Martin; Maxera, Pavel; Seitl, Martin; Dominik, Tomáš; Semela, Marek; Šucha, Matúš; Kolář, Radim · 2020 · OpenAlex-citations

DOI: 10.1109/tits.2020.2977762

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This review paper addresses the critical safety issue of driver fatigue and stress, which significantly contribute to traffic accidents and associated socio-economic costs. The authors aim to provide a comprehensive state-of-the-art overview of multimodal detection approaches, focusing on the specific features extracted from biological, vehicle, and video data. The study is motivated by the need to improve the reliability of existing alert systems and support the development of automated driving technologies that require accurate mental state assessment. The paper synthesizes existing literature rather than presenting new experimental data. It categorizes available datasets into driving-specific repositories (e.g., DRIVEDB, Warwick-JLR) and general physiological databases (e.g., DEAP, ASCERTAIN), noting the lack of a single comprehensive database containing complex driver, vehicle, and annotated stress/fatigue data. The authors detail the standard detection pipeline, which includes data recording, preprocessing, feature extraction, feature selection, and classification. They analyze three primary data sources: vehicle data, physiological signals, and driver behavior. Regarding findings, the review highlights that vehicle-based indicators, such as steering wheel angle, lane deviation, and yaw rate, are widely used but suffer from inconsistencies due to road geometry, weather conditions, and inter-individual variability. Consequently, vehicle-only measures are deemed unreliable for standalone fatigue prediction. In contrast, physiological features are identified as the most robust and reliable indicators because they are directly linked to the driver’s physical and psychological state and are less susceptible to environmental artifacts. The paper extensively details specific physiological metrics, including heart rate variability (HRV) features derived from electrocardiography (ECG) and photoplethysmography (PPG), respiratory rates, electrodermal activity (EDA), brain activity, and body temperature. For instance, HRV features in time, frequency, and non-linear domains are shown to effectively distinguish stress states, with the LF/HF ratio increasing during stress. The significance of this work lies in its conclusion that multimodal fusion is essential for accurate detection. While physiological signals offer high reliability, they can be invasive or uncomfortable; vehicle data offers contactless monitoring but lacks specificity. The authors advocate for combining physiological and behavioral parameters to mitigate individual limitations and improve system accuracy. This comprehensive mapping of features and datasets provides a foundational reference for researchers developing robust, real-time driver monitoring systems for both current and future autonomous vehicles.

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-prismaquant on 2026-06-26; verification: verified.

Topics

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

• drowsiness detection algorithms
• stress driving
• distraction detection algorithms

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