An Intelligent Vision-Based Feeding and Hydration System for Small Domestic Rodents

Abstract

This paper introduces a cost-effective intelligent feeding and hydration platform for small domestic rodents. The system integrates computer vision, mechatronic actuation, and Internet of Things (IoT) connectivity within a unified perception–decision–action framework. A YOLOv8-S model identifies Russian hamsters, Syrian hamsters, and guinea pigs, dispensing species-specific portions of 5 g, 8 g, or 10 g. The ESP32 microcontroller manages a 3D-printed rotary dosing wheel actuated by an SG90 servomotor, verifies dispensed mass using a 1 kg load cell with an HX711 amplifier, and prevents empty-container operation using an infrared sensor. Hydration is controlled by a water-level sensor and dual submersible pumps for filling and drainage. Data and alerts are transmitted via MQTT and visualised through a Blynk mobile interface, enabling real-time monitoring and manual override. Experimental results from 60–40 trial batches demonstrate stable sensing and communication. The vision model achieves 99.667% test accuracy on 82 images from a balanced 819-image dataset, supporting reliable autonomous care.