Is your heating radiator working? This system uses edge AI to find out

Heating isn’t just about comfort in very cold regions in the world, it is absolutely critical. If your home’s heat fails when you’re away, you might get back to find burst pipes. That’s especially problematic for large apartment buildings that rely on central boilers and radiators in the units. Manivannan’s Arduino-based edge solution leverages AI to detect those radiator failures and raise an alarm.

When we talk about technology “on the edge,” we mean that the physical device does its own computing and processing onsite. That is in contrast to cloud-based computing that requires an internet connection and data transfer, which has ongoing costs and the potential for downtime if the connection fails. Manivannan’s setup is designed around an Arduino Portenta H7 running an Edge Impulse ML model, along with a microphone and a DHT11 temperature and humidity sensor.

The system monitors the temperature in a room and listens for the sound of the boiler running. If it hears the boiler running for a set period of time (like 30 minutes) and doesn’t see a corresponding rise in temperature, it concludes that something is wrong. The Edge Impulse model does the listening, observing the audio data processed with a Fast Fourier Transform (FFT).

In a real-world scenario, the Arduino would notify a responsible party in the event of a detected heating anomaly. It could do that by connected means, such as push notifications, but it could also do so with a simple buzzer or even a flashing LED.