SafeRoute AI

Inspiration

What it doesSafeRoute

Inspiration

Road accidents often increase during adverse weather conditions such as heavy rain, storms, fog, and snowfall. While existing navigation applications focus mainly on the fastest or shortest route, they rarely consider weather-related risks. We wanted to build a solution that prioritizes traveler safety by combining real-time weather data with Artificial Intelligence to recommend the safest route.

What it does

SafeRoute is an AI-powered weather-aware navigation system that generates up to three alternative routes between a source and destination. It collects real-time weather information at multiple points along each route and uses a trained Machine Learning model to predict the safety risk level.

The application provides:

Multiple alternative routes
Real-time weather conditions
Route risk score (Safe, Low, Moderate, High)
Confidence score
Risk explanation
Safety recommendations

This helps users make informed travel decisions instead of choosing only the fastest route.

How we built it

We developed the project using a full-stack architecture.

Frontend

React 19
Vite
React Leaflet
Axios

Backend

FastAPI
Uvicorn
Pydantic

APIs

OpenRouteService for route generation
Open-Meteo for real-time weather
Nominatim for geocoding

Machine Learning

We trained multiple classification models, including:

LightGBM
Random Forest
XGBoost
CatBoost
ExtraTrees
Gradient Boosting

The best-performing model was automatically selected based on the F1-Macro score and used for route risk prediction.

How it works

The user enters the source and destination.
Nominatim converts the locations into coordinates.
OpenRouteService generates up to three alternative routes.
Four evenly spaced points are selected on each route.
Open-Meteo provides real-time weather for each point.
Weather data is sent to the FastAPI backend.
The Machine Learning model predicts the risk for every segment.
The overall route risk is calculated.
The safest route and driving recommendations are shown to the user.

Challenges we faced

Integrating multiple APIs into one workflow.
Collecting weather data efficiently for different route points.
Training and comparing multiple Machine Learning models.
Designing an effective route risk calculation.
Synchronizing frontend map visualization with backend predictions.
Handling API failures with a fallback risk scoring method.

What we learned

This project helped us gain practical experience in:

Full-stack web development
REST API integration
Machine Learning model training and evaluation
Real-time data processing
Interactive map visualization
Building AI-powered applications

What's next

We plan to improve SafeRoute by adding:

Live traffic analysis
Accident prediction
Flood and road closure detection
Driver-specific recommendations
Mobile application support
Emergency alert integration
Personalized route preferences

Impact

SafeRoute transforms traditional navigation into a safety-first navigation system by combining AI, weather analysis, and real-time routing. Instead of simply guiding users to the fastest destination, it helps them choose the safest route for their journey.