AI-Driven Real-Time Driver Monitoring and Intelligent Safety Intervention Using Deep Learning Models

Abstract

Worldwide, driver weariness and distraction are major causes of traffic accidents. This study describes an AI-driven Driver Monitoring System (DMS) that detects tiredness, distraction, and risky driving behaviours in real time using computer vision, deep learning, and sensor fusion. The suggested system calculates a risk probability index by combining an infrared camera in the cabin with a steering angle sensor and optional physiological inputs. Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTMs) are combined in a multi-stage deep learning framework to analyse temporal behaviour. Real-time intervention mechanisms, such as vibration feedback, auditory alarms, and simulated braking control, are demonstrated in a hardware simulation prototype that uses embedded edge devices. Advanced Driver Assistance Systems (ADAS) can locate objects with high accuracy and minimal latency, according to experimental testing [1].