UNIFIED FRAMEWORK FOR REAL-TIME MULTI-FACTOR RISK-AWARE ROUTE OPTIMIZATION USING MACHINE LEARNING

Abstract

In the era of smart transportation and increasing road traffic, traveler safety is a critical priority. Traditional navigation systems often minimize distance or time, overlooking contextual safety factors. This paper presents a unified, machine learning-based framework for multi-factor risk assessment and safe route optimization. The system integrates static parameters (vehicle type, user demographics, lighting, road type, group size) and dynamic parameters (real-time weather, traffic, surface condition, time). It gathers data from Google Maps, OpenStreetMap, OpenWeatherMap, and TomTom, processes it through a modular pipeline, and assigns risk values to route segments. Gradient Boosting, Random Forest, and Decision Tree models predict cumulative risk scores to recommend the safest route. A case study between Maragathapuram and Parvathipuram in Tamil Nadu validates the framework. The experimental results show Gradient Boosting provides the highest prediction accuracy. The system provides real-time, personalized, and data-driven travel recommendations, enhancing intelligent transportation with a focus on safety.