Abstract

Most public toilets are cleaned on a timetable. Nobody counts how many people actually used them, nobody notices the water tank running low until it is empty, and the air inside can turn hot and damp for hours without anyone knowing. We built a small prototype to see how much of this can be handled with cheap hardware. An ESP32 reads two infrared sensors across the doorway, an ultrasonic ranger above the tank, a flow sensor on the supply line and a DHT11. A relay board runs a pump when the tank is low and a fan when the climate drifts out of range. Readings go over Wi-Fi to a Supabase table, and a browser dashboard shows several locations along with a rough hygiene score. The hardware sits in a cardboard cubicle; the same logic was also run in the Wokwi simulator, and timestamped rows were seen arriving in the cloud table. The hygiene score comes from fixed rules. A Random Forest classifier is described as a later step, but none has been trained, and we report no accuracy figure for it. Throughout, we try to be clear about what each of these inexpensive sensors can and cannot tell us.

Keywords

How to Cite This Article

Rights