Built for Hack2Skill Solution Challenge 2026 – Build with AI Track
RouteMindAI is an AI-powered predictive logistics intelligence platform that helps fleet managers anticipate delivery delays before dispatch, instead of reacting after disruptions occur. By combining route geometry, micro-weather sampling, and Google Gemini 2.5 Flash reasoning, the system generates interpretable delay-risk predictions and recommends actionable alternatives.
This project demonstrates how Generative AI + Geospatial APIs + Weather Intelligence can transform traditional logistics workflows into proactive, data-driven planning systems.
| Resource | Link |
|---|---|
| 🚀 Live Web App | View Live App |
| 🎥 Demo Video | Watch on Google Drive |
| 📊 Pitch Deck | View Slides |
| 📦 GitHub Repository | View Source Code |
Users enter shipment source and destination cities.
This triggers:
- route geometry extraction
- weather sampling along the corridor
- Gemini-based delay risk prediction pipeline
Displays AI-generated logistics insights:
Includes:
- Risk classification (Low / Medium / High)
- Delay probability percentage
- Weather-based explanation
- Suggested alternate routing strategy
Provides enterprise-style analytics such as:
- Average delay probability
- Most common disruption type
- High-risk corridor frequency tracking
Helps organizations identify long-term disruption hotspots.
Interactive map view displays:
- shipment origin
- destination
- geographic corridor overview
Supports spatial understanding of predicted risk zones.
Modern logistics operations still depend heavily on reactive planning.
Fleet operators typically:
- dispatch shipments based on static route estimates
- rely on incomplete weather insights
- detect risks only after vehicles are already in transit
- suffer delays, fuel losses, and supply-chain disruption
This leads to:
❌ missed delivery windows
❌ increased operational costs
❌ poor customer experience
❌ inefficient route planning decisions
RouteMindAI introduces an AI-driven predictive route-risk intelligence system that:
✅ samples weather conditions along the actual shipment corridor
✅ evaluates route geometry using geospatial APIs
✅ analyzes structured climate signals using Gemini 2.5 Flash
✅ generates interpretable risk probability scores
✅ recommends safer alternate logistics strategies
Instead of reacting to delays, logistics teams can now prevent them proactively.
RouteMindAI is designed for:
- Fleet managers
- Supply chain analysts
- Logistics startups
- Transport aggregators
- Warehouse dispatch coordinators
- Enterprise logistics planning teams
The system automatically:
- converts source & destination into coordinates
- generates route polylines
- extracts route nodes
- samples geographic checkpoints along the corridor
Powered by:
OpenRouteService Directions API
Instead of city-level forecasts, RouteMindAI performs:
📍 node-level weather intelligence collection
Weather parameters sampled include:
- rainfall probability
- cloud coverage
- storm intensity
- humidity conditions
This produces route-specific climate intelligence, not generic forecasts.
Powered by:
OpenWeatherMap API
Structured weather + route data is converted into a JSON payload and analyzed using:
Google Gemini 2.5 Flash
Gemini produces:
- delay probability percentage
- risk classification (Low / Medium / High)
- explanation of risk factors
- corridor-level insights
- alternate route suggestions
This transforms raw environmental signals into actionable logistics intelligence.
All predictions are stored inside:
Firebase Firestore
This enables:
📊 historical analytics
📊 route-risk trend tracking
📊 repeated delay corridor detection
📊 enterprise planning support
Organizations can identify long-term disruption hotspots.
Below is the intelligence pipeline used inside RouteMindAI:
User Input
↓
Geocoding (Source → Destination)
↓
Polyline Route Extraction
↓
Route Node Sampling
↓
Weather Sampling at Each Node
↓
Structured JSON Risk Payload
↓
Gemini 2.5 Flash Analysis
↓
Delay Probability + Explanation
↓
Firestore Storage
↓
Analytics Dashboard Visualization
Flutter Web (Dart)
Chosen because:
- single codebase supports mobile + web
- fast UI rendering
- scalable architecture
Firebase Platform
Includes:
- Firebase Hosting
- Cloud Firestore
- Secure API integration
Benefits:
- real-time sync
- scalable architecture
- zero-server deployment
Google Gemini 2.5 Flash
Used for:
- structured reasoning
- probability estimation
- corridor-risk interpretation
- alternate route suggestions
Gemini acts as a virtual logistics analyst.
| API | Purpose |
|---|---|
| OpenRouteService | Geocoding + Route Polyline |
| OpenWeatherMap | Micro-weather intelligence |
| Gemini API | Predictive reasoning engine |
User enters route
↓
OpenRouteService generates polyline
↓
System extracts sample nodes
↓
Weather fetched per node
↓
Structured JSON generated
↓
Gemini analyzes disruption probability
↓
Prediction saved in Firestore
↓
Dashboard displays insights
Example prediction:
Route: Pune → Mumbai
Risk Level: HIGH
Delay Probability: 68%
Reason:
Heavy rainfall clusters detected near Lonavala corridor
Alternate Suggested:
NH160 corridor route recommended
This makes predictions interpretable and operationally useful.
Follow these steps to run the project locally.
git clone https://github.com/BhaveshV23/RouteMindAI.git
cd RouteMindAI
flutter pub get
Create a file named:
.env
Inside project root directory.
Add the following:
GEMINI_API_KEY=your_api_key
ORS_API_KEY=your_api_key
OPENWEATHER_API_KEY=your_api_key
Run on Chrome:
flutter run --dart-define-from-file=.env -d chrome
RouteMindAI is deployed using:
Firebase Hosting
Deployment steps:
flutter build web
firebase deploy
RouteMindAI follows best practices:
✅ API keys stored in .env
✅ keys excluded from GitHub
✅ Firebase-managed hosting
✅ structured JSON prompts for safe AI usage
Upcoming enhancements planned:
- traffic congestion prediction integration
- historical delay ML training dataset
- fleet-scale batch route evaluation
- live vehicle tracking overlay
- enterprise dashboard analytics
- mobile application release
Bhavesh Vadnere
Information Technology Engineering Student
GitHub: https://github.com/BhaveshV23
Linkedin: https://linkedin.com/in/bhavesh-vadnere
If you found this useful:
⭐ Star the repository
🍴 Fork the project
📢 Share feedback
🤝 Collaborate on improvements
This project is created for educational and hackathon demonstration purposes.
Production adaptation may require:
- enterprise weather APIs
- fleet telemetry integration
- compliance alignment
"From reactive logistics to predictive intelligence powered by AI."