BUILT FOR INTERNATIONAL HACKATHON · SMART SAFETY TRACK

The helmet that sees
danger before it happens.

MineGuard AI fuses gas, thermal, motion and proximity telemetry from a wearable helmet unit into a single real‑time AI risk score — predicting hazards, guiding prevention, and triggering protection the moment a worker needs it most.

4
Fused Sensor Inputs
0–100
AI Risk Score
3
Response Stages
<1s
Alert Latency*
The Problem

Underground hazards build up invisibly

Toxic gas pockets, heat stress, unstable footing and roof collapse rarely announce themselves. By the time a worker notices, the safety window has often already closed — and traditional fixed gas detectors can't track a moving miner or catch a fall.

The Solution

Predict → Prevent → Protect

One continuous AI safety loop, running on every helmet, every second.

🔮

PREDICT

Fuses gas, temperature, humidity and motion trends into a live 0–100 risk score, forecasting hazard escalation minutes before thresholds are breached.

🛑

PREVENT

Pushes real‑time warnings and recommended actions to the worker and supervisor — reroute, ventilate, or evacuate — before conditions turn critical.

🛡️

PROTECT

MPU6050‑based fall detection instantly alerts the control room with the worker's last known zone, dispatching rescue without waiting for a call.

Core Features

Five sensing capabilities, one helmet

Every icon on the MineGuard AI badge is a live capability in this prototype.

🧪

Gas Detection

Continuous CH₄‑equivalent ppm sensing with rising‑trend alerts.

🧠

Risk Prediction

AI fuses all sensors into one composite 0–100 risk score.

🌡️

Temp & Humidity

Tracks heat stress conditions around the worker in real time.

⚠️

Collision Alert

Ultrasonic distance sensing warns of roof, wall or vehicle proximity.

📍

Real‑Time Location

Live zone tracking pinpoints exactly where each worker is underground.

How It Works

From helmet sensor to control‑room alert

📡
Sensor Fusion
Gas · Temp/Humidity · MPU6050 · Distance
→
🧠
AI Risk Engine
Composite 0–100 score
→
🗺️
Zone Mapping
Live worker + zone risk
→
🚨
Alert & Response
Worker + control room
Tech Stack

Built to be demoed — and deployed

ESP32 / Arduino MQ‑2 / MQ‑135 Gas Sensor DHT22 Temp + Humidity MPU6050 IMU Ultrasonic / IR Distance Python + Streamlit Dashboard Rule‑based + ML Risk Scoring MQTT / WiFi Telemetry
Why It Matters

Designed around the worker, not just the sensor

Per‑worker, not per‑zone
Risk travels with the miner, not a fixed post — so the system knows exactly who is exposed and where.
Minutes of warning, not seconds
Trend‑based prediction gives supervisors time to reroute or ventilate before a threshold is even crossed.
No one has to press a button
Fall detection triggers the emergency alert automatically — critical when a worker is unconscious or unreachable.
Built on low‑cost, available hardware
Every component is off‑the‑shelf, keeping the path from prototype to pilot short and affordable.
Validation

Tested against real mining‑safety scenarios

Simulated field scenarios
Sensor inputs were modeled on real underground‑mine failure patterns — slow gas buildup, heat/humidity spikes, and impact‑style falls — not random test data.
Four validated demo states
Safe → Gas Rising → Critical → Fall/Emergency were each run end‑to‑end through sensor fusion, risk scoring, and alert triggering to confirm the pipeline behaves correctly at every stage.
Hardware path already mapped
ESP32 + MQ‑2/MQ‑135 + DHT22 + MPU6050 wiring and MQTT telemetry are architected and ready for a physical pilot — the next step after this launch phase.
Launch & Feedback

Publicly deployed and open for testing

Live, public, no login required
This is not a local prototype — MineGuard AI is deployed and reachable by anyone with the link, right now, for real user testing.
We want your feedback
Tried the dashboard? Tell us what worked, what confused you, and what you'd want to see in a real deployment.

📝 Share feedback (2‑min form) →
Team

Built by

Bhaskar Nath Goswami
Solo builder — ECE background.

📧 Contact: bhaskarnathgoswami27@gmail.com
🔗 LinkedIn: bhaskar-goswami-336186412
Live Demo  ▸
Helmet Unit HLM‑0417 · J. Meena
Link Telemetry OK
Local Time --:--:--
Gas Concentration NORMAL
12ppm (CH₄ eq.)
0Danger 150+
Temp / Humidity
27°C
45%RH
AmbientComfortable
⚠️ Collision Alert CLEAR
180cm to obstruction
Roof / wall / vehicle clearanceDanger <60cm
MPU6050 Orientation
2°
UPRIGHT
AI Composite Risk Score t = 0s
12
LOW RISK
🔮PREDICT — AI Forecast

Risk trending stable. No escalation predicted in the next 30 minutes.

Nominal
📍 Real‑Time LocationZone A
Live position 24.8100°N, 83.0200°E
Safe Elevated High risk Worker
Incident History0 events
🚨
FALL DETECTED — CONTROL ROOM ALERTED
Dispatching rescue team to Zone D · Worker HLM‑0417