DRIVE FUSION - A CONTEXT AWARE DRIVER MONITORING SYSTEM

Fatigue and distraction among drivers are the leading causes of road traffic accidents all over the world with a
considerable percentage of deaths and severe injuries every year. Conventional driver monitoring systems are based on
isolated detection methods, vehicle specific hardware or periodic alarm notifications, which restricts contextual awareness,
scalability, and user interaction. This paper introduces Drive Fusion, a multimodal driver intelligence system based on real-
time perception and contextual risk modeling along with conversational artificial intelligence to support road safety. The
system uses facial landmark tracking (based on MediaPipe) to estimate driver alertness, object detection (YOLOv8) to
observe the in-cabin distraction and surrounding traffic, and lane analysis (OpenCV) to determine driving stability. The
Google Gemini API is used to interpret structured driving metrics and produce contextual and explainable risk metrics,
whereas the ElevenLabs Conversational AI API is used to provide proactive voice-based feedback, instead of the traditional
alarm system, as an adaptive and human-centered interaction. Drive Fusion is optimized to run on consumer grade mobile
devices and converts binary classification fatigue detection to contextual risk into contextual risk intelligence to enhance real-
time responsiveness and minimize false alerts.