Real-Time Detection System of Driver Distraction Using Machine Learning

Tango, Fabio; Botta, Marco · 2013 · Crossref

DOI: 10.1109/tits.2013.2247760

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 safety issue of driver distraction, which is a leading cause of vehicle crashes and fatalities. With the increasing prevalence of In-Vehicle Information Systems (IVIS) and Partially Autonomous Driving Assistance Systems (PADAS), there is a growing need for real-time detection of driver status to adapt these systems accordingly. The authors propose a non-intrusive method for detecting visual distraction using only vehicle dynamics data, avoiding the need for eye-tracking hardware as input to the classifiers. The study aims to compare various Machine Learning (ML) models to determine the most effective approach for classifying distracted versus non-distracted driving states. The experimental design utilized a static driving simulator with 20 human subjects divided into two age groups (20–25 and 30–45 years). Distraction was induced using a Secondary Visual Research Task (SURT), requiring drivers to locate and touch a specific symbol on a side-mounted display. Vehicle dynamic data, including speed, steering angle, lateral position, and pedal positions, were collected at 20 Hz. Ground truth labels for distraction were established by an experimenter using infrared cameras to monitor eye gaze and interaction with the SURT; a driver was labeled as distracted if their eyes were off the road for at least 1.8 seconds. The authors compared four ML techniques: Support Vector Machines (SVM), Feed-Forward Neural Networks (FFNN), Layer-Recurrent Neural Networks (LRNN), and Adaptive-Network-based Fuzzy Inference Systems (ANFIS). An "intra-subject" analysis approach was employed, training individual models for each participant to account for personal driving styles. The results demonstrated that SVM outperformed all other methods, achieving the highest classification rates for the majority of subjects. While FFNN and LRNN models showed variable performance, with some subjects achieving acceptable classification rates (>80%), SVM provided more consistent and superior accuracy. The study found that intra-subject modeling was necessary, as inter-subject models previously yielded poor results (~75% accuracy) due to significant individual variations in driving behavior. SVM models frequently achieved "good" classification rates (>90%) for many participants, whereas other methods struggled to maintain high specificity and sensitivity across the dataset. The significance of this research lies in its demonstration that driver distraction can be reliably detected using non-intrusive vehicle dynamics data and ML algorithms, specifically SVM. This approach facilitates the development of adaptive IVIS and "smarter" PADAS that can mitigate distraction risks by adjusting assistance strategies based on real-time driver status. By avoiding intrusive eye-tracking sensors, the proposed system enhances user acceptability and practical implementation in vehicles. The findings support the integration of ML-based driver monitoring systems to improve road safety, particularly as vehicles become more automated and equipped with complex information 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_with_issues.

Topics

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

• distraction detection algorithms
• drowsiness detection algorithms
• visual
• gaze based attention detection
• telematics crash prediction
• situational awareness

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).

• Methodological Resource: tool software
• Theoretical Contribution: computational model, conceptual framework