FLURON
AI-powered early fire detection & rapid response system
Computer Vision · Thermal Sensing · Autonomous Navigation
THE PROBLEM
Wildfires burned over 2.6 million acres in the US in 2023 alone
Fire-related disasters cause over $21 billion in property damage annually worldwide
Most destruction happens because fires are detected too late — not because we cannot fight them
A fire caught in the first 5 minutes can be contained by a single responder
The same fire left undetected for 30 minutes becomes an uncontrollable catastrophe
The difference between a minor incident and a disaster is always time
WHY CURRENT SOLUTIONS FAIL
| Method | Cost | Speed | Coverage | Outdoor | Accessible |
|---|---|---|---|---|---|
| Smoke Detectors | Low | Fast | Indoor only | No | Yes |
| Satellite Imaging | Free access | 15 min to hours | Very wide | Yes | Data only |
| Manned Watchtowers | High staffing | Variable | Line of sight | Yes | Difficult |
| Fixed CCTV + AI | High install | Moderate | Fixed area | Yes | No |
| Enterprise Drones | $10,000-$30,000+ | Fast | Wide | Yes | No |
No existing solution is simultaneously affordable, mobile, real-time, autonomous, and accessible.
FLURON
An autonomous drone that sees fire before it spreads
01 — Patrols high-risk areas autonomously — forests, warehouses, industrial zones
02 — Detects fire and smoke in real time using a fine-tuned Vision LLM
03 — Alerts responders instantly with GPS coordinates, visual confirmation, and severity assessment
HOW IT WORKS
01
PATROL
Drone autonomously patrols predefined high-risk route
02
CAPTURE
Camera module captures continuous live video feed
03
DETECT
Fine-tuned Vision LLM analyzes frames for fire and smoke
04
ASSESS
AI evaluates severity, location, surroundings, and risk level
05
ALERT
GPS-tagged alert with contextual report sent to responders
06
RESPOND
(Future) CO2 suppression deployed to buy critical time
WHY VISION LLM
| Factor | Traditional Detection (YOLO) | Fluron (Fine-Tuned Vision LLM) |
|---|---|---|
| Output | "Fire detected, 87% confidence" | "Small ground fire near wooden structure, early stage, moderate severity, immediate attention needed" |
| New Environments | Requires retraining | Generalizes with broad visual understanding |
| False Positives | Triggered by sunlight, orange objects, steam | Reasons about full context to distinguish real fire |
| Severity Assessment | Not capable without extra systems | Built-in as natural output |
| Reporting | Requires separate processing | Generates actionable natural language reports directly |
Fluron does not just detect fire. It understands the situation.
WHAT POWERS FLURON
| Component | Role |
|---|---|
| Quadcopter Frame | Aerial platform for autonomous patrol |
| Camera Module | Captures live video for AI analysis |
| Vision LLM (Fine-Tuned) | Core intelligence — detects fire with contextual understanding |
| Flight Controller (Pixhawk) | Manages autonomous navigation and stabilization |
| GPS Module | Geo-tags incidents and guides patrol routes |
| Edge Compute Board | Runs AI inference on-device for real-time speed |
| Communication Module | Transmits alerts and telemetry to ground station |
| CO2 System (Planned) | Early-stage fire suppression for future phase |
WHY FLURON WINS
| Factor | Enterprise Drones | Fixed Systems | Fluron |
|---|---|---|---|
| Cost | $10,000-$30,000+ | $5,000+ install | Under $2,000 target |
| Mobility | Yes | No | Yes |
| Real-Time Detection | Yes | Moderate | Yes |
| Contextual AI | Basic | Basic | Advanced Vision LLM |
| Accessible to Small Communities | No | No | Yes — designed for it |
| Autonomous | Partial | No | Fully autonomous patrol |
Enterprise capability at a fraction of the cost. Built for the communities that need it most.
WHY THIS MATTERS TO US
We started this project after watching news coverage of wildfires destroying entire communities — families evacuated with minutes of warning, homes reduced to ash, ecosystems wiped out. And we kept asking the same question: we have drones, we have cameras, we have AI — why are we still relying on someone happening to see smoke from miles away?
We are high school students. We do not have a lab or a massive budget. But we have the engineering skills, the drive, and the belief that detecting a fire five minutes earlier can save lives, homes, and entire forests.
That conviction is what built Fluron.
WHAT'S NEXT
Research, system design, and AI prototyping
Working detection drone with live Vision LLM — Next 3 months
Pilot testing with local fire departments or facilities
CO2 suppression module integration and testing
Field deployment and community partnerships
WHAT WE NEED
Funding
$1,500 takes Fluron from concept to a fully functional detection prototype ready for field testing.
| Item | Cost |
|---|---|
| Commercial drone frame + flight controller | $400-600 |
| Upgraded camera module | $50-100 |
| Edge compute board (Jetson Orin Nano) | $500-700 |
| Thermal sensor module | $100-200 |
| Testing supplies and safety equipment | $50-100 |
Mentorship
Connections to mentors in drone engineering, fire safety, and computer vision to guide our development.
Feedback
Honest feedback from the judges on our approach, our roadmap, and how we can improve.
FLURON
Early Eyes Before the Fire Spreads
Team Fluron
samiryzy@gmail.com