The system uses a Raspberry Pi to control a Yaskawa motor driver via Modbus, automating daily fish feeding based on various parameters.
The initial PLC-based system was replaced with a more flexible Raspberry Pi and eventually a powerful mini PC to overcome debugging and control limitations.
Food is transported from silos, measured by weights, and moved via a chain conveyor system to individual feeder hoppers.
The architecture is split between the AquaFeeder for individual tank control and the AquaMaster for overall logistics and UI.
The feeder's Node-RED flows handle configuration, statistics, and an automated recovery process for SD card failures using its MAC address.
The AquaMaster uses a queue worker in Node-RED to manage empty feeders, control weights, and communicate with the Blazor UI via an embedded MQTT broker.
The development environment is containerized using Docker and integrated with Azure DevOps for CI/CD, with a simulator for testing the Blazor UI.
Node-RED provides stability and rapid visual debugging for IoT, but challenges like fleet updates and comprehensive testing remain.
Lessons learned include prioritizing edge-first control, separating responsibilities for fallback, and standardizing deployments with containers.
Jobs that call for the skills explored in this talk.