Inspiration
Wildfires and large-scale disasters are increasing every year. Although drones and robots are being used for surveillance, most systems lack intelligent and explainable decision guidance.
Robotics without decision intelligence leads to inefficient and unsafe deployments.
This motivated the creation of FireGuard Autonomous Disaster Intelligence Platform, focusing on combining AI-based risk assessment with robotics mission planning.
What it does
FireGuard predicts disaster risk using environmental data and converts the risk into robot deployment recommendations.
LOW → No deployment MEDIUM → Ground robot patrol HIGH → Aerial drone surveillance
The system works as a decision-support layer, not an autonomous controller.
How we built it
The platform was built as a full-stack AI system.
Random Forest model for risk prediction
Feature importance for explainability
Flask REST API for inference
Web dashboard for operator interaction
Risk estimation is based on an ensemble approach:
1 / 𝑛 1/n
Example logic:
puts "Mission recommendation generated"
Challenges we ran into
Designing robotics logic without overclaiming autonomy
Balancing explainability and model performance
Debugging frontend and backend communication
Deploying the system with limited infrastructure
Accomplishments that we're proud of
Built an end-to-end AI + robotics decision system
Added mission planning for robotic deployment
Deployed a working ML application
Completed the project as a solo participant
What we learned
Explainability is critical for trust in robotics
Robotics systems require strong AI decision layers
Deployment challenges define real-world ML
Honest scoping improves credibility
What's next for FireGuard Autonomous Disaster Intelligence Platform
Integration with real-time drone data
Map-based mission planning
Robotics simulation testing
Expansion to other disaster scenarios
Built With
- css
- flask
- git
- gunicorn
- html
- javascript
- joblib
- machine-learning
- numpy
- open-meteo-api
- pandas
- python
- random-forest
- render-cloud
- rest-api
- scikit-learn
Log in or sign up for Devpost to join the conversation.