Canary

CAD of circuit casing
sensor labels
Rasp pi circuit

Inspiration

The Dominican Republic has some of the largest mineral reserves in North America. Mines employ 1% of the working population but are responsible for 8% of deaths. One of the main causes is scentless gases like methane and carbon monoxide. In the US, this isn't as much of an issue because of sensors, but in 3rd world countries like the Dominican Republic, these sensors are just too expensive. We will make it cheaper.

What it does

This project acts as a modern “digital canary” for mines. It continuously detects dangerous gases and environmental changes, providing early warnings before conditions become life-threatening. Simulating and mapping gas spread helps predict danger zones and enables faster, safer evacuation.

How we built it

We built the system around a Raspberry Pi, interfacing MQ-series gas sensors and a DHT11 humidity sensor through its GPIO pins. The sensors continuously collect environmental data, with thresholds used to detect hazardous conditions. A Python-based pipeline processes this data in real time, triggering alerts when unsafe levels are reached. We also integrated a depth camera to map the surrounding tunnel geometry, enabling more accurate simulations of gas diffusion and allowing us to visualize and predict how hazards spread through the environment.

Challenges We Ran Into

Integrating and interfacing with the depth camera module
Extensive circuit validation, including continuity checks, waveform analysis, and oscilloscope measurements
Designing and executing test cases to verify system functionality
Refining user experience and interface (UI/UX) for clarity and usability

Accomplishments that we're proud of

Got a real-time gas + humidity monitoring system up and running from scratch
Successfully connected all our sensors to a Raspberry Pi and built a live data pipeline that actually works
Added hazard detection with automatic alerts so it’s not just collecting data, but acting on it
Built a simulation to visualize how gas spreads through tunnels
Took it a step further by using that data to predict potential danger zones
Worked through a lot of hardware bugs and calibration issues to get reliable readings

What we learned

Sensor calibration is critical for reliable readings
Real-time systems require fast, efficient data handling
Simple sensors can power meaningful safety applications
Visualization helps turn raw data into actionable insights

What's next for Canary

Scale into a distributed network of low-cost sensors across mines, enabling real-time monitoring, predictive alerts, and faster emergency response. We’d also transition to a custom PCB to make the system smaller and cheaper, including a lighter power supply.