Driver Drowsiness Detection Based on Steering Wheel Data Applying Adaptive Neuro-Fuzzy Feature Selection.

Area of Science:

• Automotive Safety
• Artificial Intelligence
• Biomedical Engineering

Background:

• Driver drowsiness is a major cause of road accidents.
• Existing drowsiness detection systems often rely on invasive methods or complex sensor setups.
• There is a need for accurate, non-invasive driver monitoring systems.

Purpose of the Study:

• To develop a novel, non-invasive driver drowsiness detection system.
• To enhance classification accuracy by employing an effective feature selection method.
• To utilize steering wheel data for real-time drowsiness assessment.

Main Methods:

• A hybrid feature selection approach combining filter and wrapper methods.
• Utilizing an adaptive neuro-fuzzy inference system (ANFIS) for feature importance evaluation.
• Employing a support vector machine (SVM) for binary classification (drowsy/awake).
• Optimizing the system using particle swarm optimization (PSO) based on classification accuracy.

Main Results:

• The proposed feature selection method effectively identifies critical features related to drowsiness.
• The developed system achieved high accuracy in detecting driver drowsiness.
• The ANFIS-SVM hybrid model demonstrated superior performance compared to existing algorithms.

Conclusions:

• The novel feature selection technique significantly improves driver drowsiness detection accuracy.
• Steering wheel data is a viable source for non-invasive drowsiness monitoring.
• The proposed system offers a promising solution for enhancing road safety.