A multimodal driver monitoring benchmark dataset for driver modeling in assisted driving automation

Nobari, Khazar Dargahi; Bertram, Torsten · 2024 · Crossref

DOI: 10.1038/s41597-024-03137-y

archive: archived pipeline: cataloged verified

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

Summary

This paper introduces **manD 1.0**, a multimodal benchmark dataset designed to support driver modeling and monitoring in the context of assisted driving automation. The authors address the limitation of existing datasets, which often focus on isolated factors like drowsiness or distraction, by providing a comprehensive resource that captures the interplay between driver state, vehicle dynamics, and environmental conditions. The dataset aims to facilitate data-driven modeling, prediction of driver reactions, and research into motion sickness and interaction strategies. The data were collected from **50 gender-balanced participants** (aged 21–65) using a static driving simulator under controlled laboratory conditions. Participants drove through five distinct scenarios involving automation levels ranging from **SAE Level 0 (manual)** to **SAE Level 3 (conditional automation)**. The experimental design employed stratification for age and gender and partial counterbalancing to mitigate carryover effects. To capture holistic driver states, the study utilized a synchronized multi-sensor suite: an **Intel RealSense camera** for facial RGB images, a **SmartEye system** for gaze tracking, an **Empatica E4 wristband** for physiological metrics (heart rate, HRV, electrodermal activity, skin temperature), a **BIOPAC B-ALERT X10 headset** for EEG and ECG, and **BodiTrak seat-pressure sensors** for body position and activity. Additionally, vehicle data and environmental conditions were recorded, and subjective emotional states were assessed via the Differential Emotions Scale questionnaire. The resulting dataset includes synchronized environmental, vehicle, and driver state data, covering physiology, body movements, activities, gaze, and facial information. Notably, data from **11 participants who experienced motion sickness** are included separately to support specific research into this phenomenon, while the remaining 39 participants constitute the primary test group. The dataset provides a rich spectrum of driver state factors derived from established cognitive architectures (ACT-R, QN-MHP, CLARION), encompassing sensory perception, decision-making, and motor response metrics. The significance of manD 1.0 lies in its comprehensiveness and statistical reliability, offering a robust foundation for interpreting and predicting driver behavior across varying automation levels. By integrating multiple modalities and accounting for covariates like age and gender, the dataset enables researchers to model complex interactions between driver states and driving contexts. This resource supports advancements in designing safer human-machine interactions and improving the understanding of driver behavior in automated driving systems.

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 2026-06-09; verification: verified.

Topics

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

• situational awareness
• 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: dataset resource, tool software