2C3-3　Real-Time Estimation of Mental-WorkLoad for Drivers Based on Multidimensional Biosignal Analysis

SHIMIZU, Takeshi; MUKAEDA, Takayuki; SHIMA, Keisuke; MURAJI, Syu; MATSUO, Juntaro; HORIUE, Masayoshi · 2019 · Crossref

DOI: 10.5100/jje.55.2c3-3

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 addresses the critical issue of driver mental workload, a significant factor in traffic accidents, particularly as modern vehicles introduce complex user interfaces that increase cognitive load. The authors propose a real-time estimation system for driver psychological load based on the analysis of multidimensional biosignals. The motivation stems from the limitation of existing methods, which often suffer from high variability in evaluation values and lack the capability for real-time assessment. By accurately estimating mental workload, the system aims to evaluate the impact of driving tasks on drivers and potentially mitigate accident risks caused by insufficient attention resources. The proposed method involves measuring various biosignals, including electrocardiogram (ECG) and eye-tracking data, alongside vehicle operation data such as steering angle and pedal depth. These signals are segmented and processed to extract feature vectors. To handle the high dimensionality and individual variability of biosignals, the authors employ Time-series Discriminant Component Analysis (TSDCA), a probabilistic neural network capable of dimensionality reduction and time-series data identification. This model estimates the posterior probability of the driver’s state (e.g., normal vs. high-load). A regression model then maps these probabilities to mental workload scores, specifically using the NASA-TLX scale, with parameters determined via the least squares method. The experimental validation was conducted using a driving simulator with six male subjects aged 23–24. Participants performed a primary task of following a lead vehicle at 80 km/h on a straight road, combined with secondary N-Back cognitive tasks (0-Back to 3-Back) to induce varying levels of mental load. The study measured ECG, gaze information, and vehicle dynamics. Results demonstrated a strong positive correlation between the estimated and measured NASA-TLX values, with a correlation coefficient of R = 0.705 and a determination coefficient of R² = 0.4965. The estimation error rate was 9.37% for training data and 22.29% for test data. Notably, the system achieved higher accuracy in estimating high-load conditions (2-Back task), with an average error rate of 10.10%. The authors attribute lower accuracy in low-load conditions to erratic biosignal fluctuations caused by inattention or gaze deviation, whereas high-load conditions resulted in more stable biosignal indicators due to focused cognitive resource allocation. The significance of this work lies in its demonstration that real-time estimation of driver mental workload is feasible using multidimensional biosignal analysis. The system effectively identifies high-load states, which are critical for safety interventions. The authors conclude that while the current model performs well under high cognitive load, future work should focus on increasing the number of subjects and evaluation metrics, as well as analyzing the contribution rates of specific indicators to improve accuracy across all load levels and driving conditions.

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.

• workload measurement
• mental demand
• stress driving
• drowsiness 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
• Theoretical Contribution: theory or model