Sleep Detection System for Trucks: A Real-Time Multi-Modal Data-Driven AI Framework for Driver Fatigue Monitoring

Fatigue-induced driving remains one of the most severe and under-addressed threats to road safety, particularly among commercial truck drivers tasked with long, monotonous routes. In response to rising accident statistics and increasing demand for proactive driver monitoring, this research proposes a comprehensive, multi-modal fatigue detection system specifically designed for real-time deployment in heavy vehicle environments. The proposed system fuses three primary input modalities: behavioral signals (Eye Aspect Ratio – EAR), physiological indicators (Heart Rate Variability – HRV), and vehicular behavior (Steering Entropy – SE). A hybrid deep learning architecture is employed, combining Convolutional Neural Networks (CNN) for spatial feature extraction with Long Short-Term Memory (LSTM) networks to capture temporal fatigue trends. Data collected through camera systems, HRV sensors, and CAN-bus telemetry are processed in real-time to generate fatigue alerts before critical drowsiness thresholds are crossed. Evaluation of the model reveals a classification accuracy of 93%, with precision, recall, and F1-score exceeding 90% across key scenarios. Latency analysis confirms system responsiveness with an average detection time of ~120 ms, making it suitable for real-world deployment. Additionally, visual analysis techniques—including feature contribution plots, clustering views, and PCA projections—are incorporated to improve transparency and interpretability of the model’s decision-making pipeline.