Real-Time Detection of Distracted Walking Using Smartphone IMU Sensors with Personalized and Emotion-Aware Modeling

Area of Science:

• Human-Computer Interaction
• Mobile Sensing
• Machine Learning

Background:

• Pedestrian behavior analysis is crucial for urban safety and personalized health.
• Existing methods often require external hardware, limiting scalability and cost-effectiveness.
• Smartphone sensors offer a ubiquitous platform for unobtrusive behavior monitoring.

Purpose of the Study:

• To introduce GaitX, a novel system for real-time pedestrian behavior recognition using only smartphone sensors.
• To detect abnormal walking behaviors, specifically smartphone usage while walking.
• To explore the correlation between gait variability, psychological traits, and mobility analytics.

Main Methods:

• Utilized multivariate time-series features extracted from smartphone accelerometer data.
• Employed ensemble machine learning models, including XGBoost and Random Forest, for classification.
• Integrated Myers-Briggs Type Indicator (MBTI) personality profiling for psychological trait analysis.

Main Results:

• Achieved an average classification accuracy of 92.3% across 21 subjects.
• Demonstrated high precision (97.1%) in identifying distracted walking behavior.
• Revealed potential links between gait variability and distinct psychological profiles.

Conclusions:

• GaitX provides a scalable and cost-effective solution for real-time pedestrian behavior recognition.
• The system enhances mobile safety applications by detecting distracted walking.
• Findings suggest potential for emotion-aware mobility analytics and personalized health monitoring.