Vehicle Driver Monitoring through the Statistical Process Control

Area of Science:

• Automotive Engineering
• Human Factors Engineering
• Data Science

Background:

• Driver monitoring systems are crucial for road safety.
• Traditional methods often struggle with real-time detection of subtle behavioral changes.
• Integrating vehicle and driver behavior data offers a comprehensive approach to safety.

Purpose of the Study:

• To apply Statistical Process Control (SPC), specifically the Exponentially Weighted Moving Average (EWMA) method, for driver behavior monitoring.
• To develop and evaluate novel methods for lane departure detection, sudden driver movement identification, and driver fatigue detection.
• To leverage in-vehicle sensor data for enhanced driver safety analysis.

Main Methods:

• Utilized Exponentially Weighted Moving Average (EWMA) charts for anomaly detection in driver behavior.
• Developed a lane departure detection algorithm based on SPC principles.
• Implemented methods for detecting sudden steering and braking actions.
• Combined computer vision with SPC for driver fatigue identification.

Main Results:

• Lane departure detection achieved up to 84.16% accuracy under specific conditions.
• Sudden movement detection reached high accuracy, up to 94.44% for braking and 91.66% for steering.
• Driver fatigue was detected in up to 94.46% of situations.
• The proposed SPC-based methods proved effective in identifying unwanted driver actions.

Conclusions:

• Statistical Process Control (SPC) methods, particularly EWMA, are highly effective for real-time driver monitoring.
• The developed algorithms provide accurate detection of critical safety events like lane departures, sudden movements, and driver fatigue.
• In-vehicle sensor data, when analyzed with SPC, significantly enhances driver safety systems.