COACH: monitoring mechanical and digital meters

The challenge

The periodic monitoring of offline mechanical and digital meters in a household context is a tedious and error-prone process. And when hundreds of them must be checked simultaneously in industrial environments, the process can become unaffordable, expensive, and dangerous. Nevertheless, many smart city projects rely on the online availability of such measures in order to improve the power supply management.

The solution

COACH is the solution we achieved. The COACH (Choice On A Chip) project addresses this issue by delivering a machine & deep learning system capable of reading meters in few seconds regardless of their visual aspect and the operative condition of the glass (dirty, crusted, opaque, etc) acting as a human reader.
The application is deployed on a IP68 device based on RaspberryPy, where the small layout coupled with portable power supply allows the installation in any location. The simple configuration allows a quick positioning and the reading at custom time steps, while converted numerical data can be easily retrieved via SMS/WiFi or stored in persistent memory for manual USB download.